Recharging a metastable module of higher quality requires plutonium of the same quality.

Is this a bug or was it intended?

Not a bug.

Balance feedback is welcome.

I don't know how feasible it is, but you could you make it so the module is used up after x amount of crafts rather than a spoil timer.

That or increase the spoil timer at higher qualities, IMO the amount of effort that is required to get legendary is not proportional with just a doubling of the spoil timer.

I would jump on that it'd be nice to rebalance this a little bit at legendary quality. The potential for legendary T4 prod modules is one of the things that can carry on into the very late game from Cerys. But, the difficulty in producing legendary plutonium makes it pretty impractical, even if you were to only use it for exclusively your biolabs. 4 modules per lab at 20 minutes each is an awful lot of legendary plutonium. The best plan I could think of was to upcycle uranium via ammunition and then let the legendary uranium be transmuted in the normal way. Maybe I'm missing something obvious, but mass upcycling 238 via ammo is still quite miserable. I was thinking of trying the artillery shells perhaps, shipping the components from vulcanus.

Perhaps instead of 1-to-1, make some number of recharge packs per plutonium, even if the other components were expensive and required a working space casino or quality chain elsewhere? Or some other recipe that could offer an upcycling path? Perhaps an alternate method for recovering spent fuel that skips the liquid step (so you could maintain the quality through the fuel cycle)?

I haven't tried it in game, but with only the input cost of legendary iron plates it seems you can sustain legendary t4 productivity modules by reprocessing legendary uranium fuel cells.

This requires an initial amount of legendary uranium necessary to kickstart the process as well as legendary t4 prod modules for all the uranium processing machines. The loop would look like the following:

1. Create legendary uranium fuel cells in an assembly machine with 4 legendary t4 prod modules at a cost of ~0.77 U238 and ~0.37 iron plates each (includes the needed ~0.04 U235 produced through kovarex enrichment).

2. Deplete the legendary uranium fuel cells in legendary nuclear reactors. Each legendary nuclear reactor takes 80 seconds to burn a fuel cell. You will want a lot of legendary nuclear reactors burning fuel cells.

3. Reprocess the legendary depleted uranium fuel cells in a centrifuge with 2 legendary t4 prod modules producing ~1.10 U238 per depleted uranium fuel cell (profit of ~0.33 U238 per fuel cell).

4. Create the necessary replacement U235 using kovarex enrichment in a centrifuge with 2 legendary t4 prod modules.

5. Convert the extra U238 into P239 using a standard method.

Using this method, each legendary nuclear reactor would produce enough legendary P239 to sustain ~5 legendary t4 prod modules before plutonium productivity research. Some of your P239 would need to go back into the prod modules used in the loop (minimum of 8 with dedicated machines). The only input cost cost would be ~1.11 legendary iron plates per P239 before plutonium productivity research. There would also be rocket logistics costs for launching the P239 off of Cerys.

That's really clever, I totally missed using the t4's in the fuel cycle itself like that.
1.63 U238 -> 1 U235 for Kovarex
20.63 U238 -> 26.8 Fuel Cells (and thus Depleted Fuel Cells)
26.8 Depleted Cells -> 29.5 U238
So overall, 20.63 U238 becomes 29.5 U238, of course rate limited by reactors as you say. 240 Legendary reactors consumes 3 fuel cells per second, as a starting point.

Once it's jumpstarted it just runs forever so you'd probably build it as a space platform floating above Cerys, with a mini casino to make legendary iron and occasionally shipping up legendary P239 to recharge the modules. Obviously you can't make reactors on Cerys anyway, this way you're only launching up legendary P239, and dropping legendary U238 is free so you're not paying the rocket tax there.