Me and my friend have been working on a music stand light (basically a clip on light) that runs 3 F4T5s on 6v ballasts where one is a 1 lamp ballast (the other lamp's portion got removed from another thing) and a 2 lamp ballast. In the stand light though, it is going to be USB powered so they will only run on 5v which makes them dimmer but still work. I've had several ballast like these for many years and they definitely get weaker where the lamps have more trouble starting but I've never actually had one fail from normal use.

Do these ballasts ever just completely fail?

Dunno what topology exactly are these, but many single transistor inverters drive emitter-base of the transistors into breakdown (mainly during lamp starting or worse when there is no lamp but the ballast still oscillates) and over time this causes the current gain of the transistors to decrease. And the operating power of the ballast is directly linked to the current gain (at the peak collector current).
Another mechanism is degrading electrolytic capacitor on the supply input (again still assuming the single transistor flyback circuit). Its gradually increasing ESR makes it less effective to really deliver the current spikes the thing needs, mainly when the wires to it are longer (= higher inductance)
So yes, these can get "weaker" over time.
But I would guess even with this I would expect the lamps would get eaten and degrade way faster than the bellast.
Other factor is, where you get the power to it. Although the average current for a F4T5 6V ballast running at 5V (by the way that is actually the mean voltage a 6V 4-cell battery ballast would be designed for, counting on the real battery voltage under load and over life) could be the 300mA ballpark, the peak current it draws is in the 1..1.5A range. And many "USB phone chargers" are just not able to deliver that. They are designed to charge batteries, not supply pulsed loads (the DCDCs the phone may have use to operate at 2MHz and that is what the filtering is designed for, these fluorescents are running at 20..30ish kHz, so may easilly confuse the regulation in the "charger").
So are you sure it is the ballast what gets weaker over time and not the lamps or the power supplies feeding it?

Thanks for the info! I don't really know that much about electronics so I'm glad you raised these issues

So first about the current limits - I've measured the current of the ballasts I've planned to use: On average it's about 230ma per lamp with not really any peak I measured weather it be with lamp(s) out, EOL lamps, or just any different condition. I've also run the setup off of my older Apple block is a 1a limit and it didn't (seem to) have a problem with it. But it seems like the pulsing might be an issue: what does it mean if the regulation in the charger is confused?

The ballasts definitely do degrade the lamps fast if not used carefully, however I have had a few lamps on better ballasts run for quite a long time (on a lantern ballast). I have one ballast so degrade it can barely run a lamp hot cathode so I use that one for EOL ones. Interestingly, some ballasts have gotten so weak running the lamps cold doesn't even degrade them that fast and the filaments don't even break (they sputter still of course but not that much). I've noticed the decline in a few of mine over the span of like 10 years over 4x 1.5v batteries, 4x 1.2 rechargeable batteries (old and new, I only use those now), and USB (various sources including laptops, blocks, and power banks) with a mix of many different lamps.

The 230mA is an average current, it internally comes by pulses about 3x higher in peak.
The thing with USB cbargers is, they need to implement current limitation so if something attempts to draw more current than the chargers can handle, they need to limit the current to the safe maximum and let the voltage collapse. The system, along with the charger controller in the phone (assume the classic specs up to 5W), is meant to operate so the charging current is limited prefferably by the charger, as that means less losses and mainly no losses in the phone itself (where is hard to manage that heat). Now this system is designed for steady current draw (i.e. connected to the battery by the controller and just letting it accept the charge), so it is not certain if with 0.75Apk 25kHz pulses it will respond to the average value (so not respond) or if it will respond to the peak value (the 0.7..1A), so steer the voltage down even when the average current is below the limit.

Definitely a battery lantern is designed to work with the batteries at the place of the lantern, not after longer wires. That could be another factor, with longer wires you need to use larger capacitor (some 1000uF/10V low ESR type) at the place where the batteries were connected, then the longer wires will help to redirect the current pulses to that capacitor and draw smoother current from the USB, so less likely to respond erratically.

That is really confusing to me !

The wires going to lamp are 8 inches longer than they used to be now will that cause any like capacitive issues too?

The ballast is based on interrupting the current into the transformer primary by a transistor.

Now if youhave long wire, it will behave as an inductor in series.
If you are trying interrupting current in an inductor, you will find the inductor is fighting you with voltage spikes, so you find out you can't, at least not fast and not without extra stress and losses on the switching element, or without reduction of the delivered power.

Normally the batteries are few cm from the ballast, so not much inductance there, so often these ballast do not use any significant capacitor across the supply. At least the cheaper ones or the ones using larger batteries for not that large power.

But a capacitor is exactly the opposite: It tends to maintainthe voltage and fights any change with current. Of couse the total charge is limited by the capacitance, but the fast pulses do not carry too much charge each, so there is time between the pulses for the capacitor to recharge back.
So if you need to power the ballast via longer wires (because of the distance), the extra inductance will rob you of the power. But adding the capacitor directly to the ballast input, the capacitor will keep the voltage steady there, even when the ballast input current is pulsing. So the wiring and everything upstream will be isolated from those pulses.

Now your 30cm is becoming significant. Not to really kill anything, but to rob some performance, yes. So you will need a capacitor on the ballast input.

It could well be the ballasts you have do feature a capacitor, but it will be rather small (assuming the short battery wires), so with your longer wires it may get stressed more and maybe degrade faster and stop doingwhat you need it for, hnece the degrading output over time.
A bit tricky is, the degradation is not that much its capacitance itself (whatmany multimeters are able to measure), but the internal series resistance increasing (and for that you need dedicated instrument, I haven't seen any universal multimeter capable measuring ESR, only dedicated meters/testers). But if there is an electrolytic capacitor on the weakening ballast board, to replace it would be good bet even when you are not able to test it. They are cheap but cause bi headaches if faulty.