Depends on the cell. Turns out, at least from my experiments that computer lico holds a charge for a long time. Even bad cells don't necessarily lose voltage even after weeks. I have cells that will only charge to 4.16 and only discharge at .45C for 10 minutes but will hold that 4.16v for weeks. This is why IMHO the "Bleed Down Test" is an inaccurate determination of the health of a cell.
All battery types self-discharge!
Test is to determine rate and depth of self-discharge ... to eliminate dangerous cells by diagnosing abnormal self discharge.
I recommend self bleed down test only to eliminate the obviously bad cells.
My, presently, favorite developed best-fastest cell testing method.
Charge all cells equally to a voltage above preferred use voltage.
I charge 40p using modded 5V MeanWell (combine cells when of nearly equal voltage )
Keep eye, or finger, on cells, remove any that start getting warm, 40p 2600mAh = 104Ah so 30A Meanwell will not create heat while charging unless cells are bad.
Separate and allow cells to set and self discharge - the longer the better. (Minimum of several days )
Eliminate all with substantial voltage loss - keep only cells that maintain above your preferred use voltage.
I used to charge to 4.20V but have begun charging to 4.05V for certain cells and 4.15V for others.
(Different variations in formulation produce different optimal charged voltages)
With all cells at equal voltage, discharge at a measured rate.
I began using 28s2p, discharging with 2 - 60w light bulbs (~120V DC discharging 120w = 1A=1000mA per hour)
2x2600mAh cells = 5200mAh, 1000mA discharge = ~.2C
Monitor each cell voltage, remove any that fall below 3.5V (voltage will drop suddenly at this voltage, so monitor carefully) and mark time, 1000mAh capacity for each hour
Discharging for 2.5 hours(50%+ oem rated capacity), 3 hours(60%+ oem rated capacity), or, if very good cells, 4 hours(80%+ oem rated capacity). Then rating cells by residual voltage, works nicely.
If cells don't last 2.5 hours, less than 50% capacity, probably not worth building into pack? (unless large bulk pack?)
Mark rated capacity on cells-pairs. (eg "3H 3.82V" or "3000mAh + 3.82V")
Method provides a fairly accurate comparative capacity ... 56 cells capacity tested in 3 hours.
Recharge all cells to equal voltage.
Build banks of equal capacity.
Line up all cells, best to worst. Shuffle into banks.
6s = 123456654321123456654321 etc
Should provide reasonably well balanced capacity banks.
Test full pack discharge, if not perfectly balanced at deep discharge, reshuffle cells to equalize, or add cell-cells to any weak bank.
Quick and easy and reasonably accurate method to test cells.