[FIX] hr_holidays: speedup Time Off dashboard

When opening the Time Off App, the JS performs an rpc to call
hr_leave_type.get_days_all_request. Inside this method there are
some __get__ on non-stored computed fields. This triggers
a recomputation of said fields in _compute_leaves. _compute_leaves
uses _get_employees_days_per_allocation.

When there is an hr.leave.type with request_unit = 'hour', you may
get lots of validated allocations by employee. When this is the case,
_get_employees_days_per_allocation gets quite slow, especially the last
part about 'Future available leaves'.

This commit optimizes this part of _get_employees_days_per_allocation.
In this part the allocations are grouped by employees
and Intervals instances are of the form (start, stop, records). This means
that for a given employee_id, all the records of one Intervals will have
the same start and stop values. This allows us to move the call to
_get_work_days_data_batch outside of the
for future_allocation_interval in future_allocation_intervals._items loop.
Because _get_work_days_data_batch is quite expansive, moving it out
grealty speeds up the _get_employees_days_per_allocation method.

Example speedup: in a database with 50 validated allocations for
the same hr.leave.type and employee_id, the timing of
_get_employees_days_per_allocation 7.39s -> 158ms

opw-3123457

closes odoo/odoo#116217

Signed-off-by: Kevin Baptiste <kba@odoo.com>

addons/hr_holidays/models/hr_leave_type.py

@@ -350,17 +350,17 @@ class HolidaysType(models.Model):
                 for future_allocation_interval in future_allocation_intervals._items:
                     if future_allocation_interval[0].date() > search_date:
                         continue
+                    employee_quantity_available = future_allocation_interval[2].employee_id._get_work_days_data_batch(
+                        future_allocation_interval[0],
+                        future_allocation_interval[1],
+                        compute_leaves=False,
+                        domain=company_domain)[employee_id]
                     for allocation in future_allocation_interval[2]:
                         if not allocation.active or allocation.date_from > search_date:
                             continue
                         days_consumed = allocations_days_consumed[employee_id][holiday_status_id][allocation]
                         if future_allocation_interval[1] != fields.datetime.combine(date, time.max) + timedelta(days=5*365):
-                            # Compute the remaining number of days/hours in the allocation only if it has an end date
-                            quantity_available = allocation.employee_id._get_work_days_data_batch(
-                                future_allocation_interval[0],
-                                future_allocation_interval[1],
-                                compute_leaves=False,
-                                domain=company_domain)[employee_id]
+                            quantity_available = employee_quantity_available
                         else:
                             # If no end date to the allocation, consider the number of days remaining as infinite
                             quantity_available = {'days': float('inf'), 'hours': float('inf')}