[IMP] hr_payroll: Improve benefits conflict performance

This commit changes the way benefits overlaps are detected.
Currently, all benefits are read from db and the
overlap detection is done in Python.

After this commit, use a single sql query to
only retreive overlapping benefits.
The query takes advantage of the postgresql
range type `tsrange` and the intersection operator

Performance improvements:
Tests with 500 employees and benefits generated for
15 months (~300000 records in the table)
Conflict detection over 1 month (~20000 records)
=> ~5X speedup (from more than 2s to less than 0.5s)

0 files changed, 0 insertions, 0 deletions