Purpose of this commit
======================
The Burndown Chart report was very slow on big databases as it was not
possible for Postgresql to optimize the query as it was based on a view
that was using several generate series.
This commit aims to improve the performance by injecting the constraints
at a lower level than it was in the past, lowering the amount of data
processed in the higher level of the query.
/!\ Important note
------------------
Overwriting the `read_group_raw` is really not a good practice and should
be avoided in most case. If you fall on this implementation by grepping
the source code, please be advised that this is not the right way of doing
things.
Implementation details
----------------------
- The report is now run by generating the `SQL` that is executed by the
`read_group_raw`. This allows inserting `SQL` constraints at a lower
level and simnifically improves performance. As there is no other way
to do it, the code is unfortunately a modified copy of the actual
`read_group_raw`.
- The pivot view has been removed as it had no meaning and was creating
confusing data.
- The `Group By` menu has been limited to `stage_id` and `date` as bringing
more data trough the different `GROUP BY` statements up to the higher level
is costly. Further more, additional `Group By` did not bring added value
as the Chart was less readable.
- The JS code has been adapted in order to force a group by both `stage_id`
and `date` so that the date displayed is always making sense.
- The sort ascending and descending options have been removed as creating
confusing data.
- The compare with previous period has also been removed as the chart only
really make sense when seen chronologically.
- A lot of tests have been added in order to ensure that changes that would
be harmful for the report will trigger test fails.
task-2845729
Co-authored-by: Yannick Tivisse <yti@odoo.com>
