First of all, you'll need to charge the cells individually. There's no way you can balance them, neither in series, nor in parallel.
Second, I recommend a controlled fast charge with preformation, 1...2 C fast charge and final top-off charge. This is the gentlest and optimum way to bring cells back. And they won't even get warm (a warm cell is a sign of bad charge management).
The U240xB series of ICs from Telefunken/Temic were designed spcifically for this, but are obsolete. But replicating the charging algorithm in a microcontroller is no problem.
https://pe2bz.philpem.me.uk/pdf%20on%20typenumber/A-C/BATTERY.pdf
See page 19.
Sorry but I have to disagree on that one....
Firstly, there is no indication that the OP's cells are fast charge capable, certainly by the lime you get into the realms of 1C or 2C!
Any NiCd will support C/10 charging. A 14 hour charge will ensure that all cells reach full charge, with an allowance for charging efficiency.
Secondly, There's no harm in charging a bunch of cells in series. With a constant current charge they will obviously all receive the same C/10 charge. Given that the OP has a 1000 of them, he's not going to want to charge them all individually. It would be a relatively cheap and simple job to arrange some series strings with LM317s as current sources (with appropriate attention to dissipation/heatsinking and voltage limit (starting from a discharged stack).
Note: I'm not suggesting that the cells wouldn't need matching for substantial discharge current in a series battery pack, simply that there's nothing wrong with charging in series strings at C/10 for the appropriate time. As amspire indicated, NiCds are highly resistant to overcharge at a C/10 rate. There would be no harm in extending a test charge to 24 hours.