[IMP] base: make recompute faster

When we read data of one record, the method `recompute` (`_fetch_field`
-> `_read` -> `flush` -> `recompute`) can take more than 40 % of the
time of the `_read`, due to a huge number of recordset creation (from
`records_to_compute` and `records & recs`).  Avoid that waste of time by
postponing the test on records.

For example, on a database with modules crm, mrp, purchase, website,
sale_management, reading the prefetchable fields of the current company
took 1.45 ms ± 60.3 µs, and now takes 1.17 ms ± 110 µs (more than 20%
speedup).

closes odoo/odoo#83818

Related: odoo/enterprise#24645
Signed-off-by: Raphael Collet <rco@odoo.com>

odoo/models.py

@@ -6141,9 +6141,9 @@ Fields:
         """
         def process(field):
             recs = self.env.records_to_compute(field)
-            if not recs:
+            if (not recs) or (records is not None and not (records & recs)):
                 return
-            if field.compute and field.store:
+            if field.store:
                 # do not force recomputation on new records; those will be
                 # recomputed by accessing the field on the records
                 recs = recs.filtered('id')
@@ -6166,18 +6166,11 @@ Fields:
             for field in list(self.env.fields_to_compute()):
                 process(field)
         else:
-            fields = [self._fields[fname] for fname in fnames]
-
-            # check whether any 'records' must be computed
-            if records is not None and not any(
-                records & self.env.records_to_compute(field)
-                for field in fields
-            ):
-                return
-
             # recompute the given fields on self's model
-            for field in fields:
-                process(field)
+            for fname in fnames:
+                field = self._fields[fname]
+                if field.compute:
+                    process(field)
 
     #
     # Generic onchange method