Fix big data in all-vs-all mode
In all-vs-all mode (only target is given), no split is done. Probably for that reason, we need too much memory and time...
- Limit all-vs-all mode to small or medium analysis (done in 1/2 h)
- Split target (so target and query [the same as the target] will be splitted... We must change the code to take that in account: needs more work (1/2 day)