[IMP] account: dashboard performances

As the number of `account.move`, `account.move.line`,
`account.bank.statement.line` and `account.journal` is growing, the
accounting dashboard, which is the entry point of the app, gets slower
and slower.

There are multiple issues being addressed in this commit:
* The data for each journal is computed journal by journal. This means
  that the number of queries run increases linearly with the number of
  journals. While the boilerplate around running multiple queries is
  negligible compared to the running time of the queries in this case,
  some queries take as much time to run for one journal or for many.
  To improve this, all the queries are now batched. This has been done
  by refactoring the code; all these functions are now called on as many
  records as needed[^1]:
  - `_get_journal_bank_account_balance`
  - `_get_journal_outstanding_payments_account_balance`
  - `_get_last_bank_statement`
  - `get_line_graph_datas`
  - `get_bar_graph_datas`
  - `get_journal_dashboard_datas`
* The gap detection and the entries' count are computed fields
  (`has_sequence_holes` and `entries_count` respectively). We don't need
  to display/compute these fields for all types of journals, but since
  they were mentioned by using a `<field/>` node in the view, they were
  computed for all journals displayed. Instead of using the `<field/>`
  node, we are now setting the value in the `kanban_dashboard` field.
* Documents in foreign currencies on journals in foreign currencies need
  to get the rate in order to be aggregated in the journal's currency.
  There are 3 cases:
  - Document in journal's currency
  - Company's currency is the same as the journal's
  - Document, company and journal have 3 different currencies
  Before this commit, the second case will still fetch the daily rate
  for the document in order to do the conversion, but we actually
  already know the conversion; it is stored on the document.

Benchmark
=========

On a `populate` database with:
- 4 `res.company` (with accounting enabled)
- 45 `account.journal`
- 19k `account.move`
- 140k `account.move.line`
- 4k `account.bank.statement.line`
- 4 `account.bank.statement`

                            | Query count | Query time | Remaining time
--------------------------- | ----------- | ---------- | --------------
Before fix                  |         279 |     0.333s |         0.375s
After fix (without update)  |          40 |     0.120s |         0.170s
After fix (with update[^2]) |          38 |     0.100s |         0.170s

Note that the currency conversion was disabled because the populate
database doesn't represent a realistic dataset regarding this. Disabling
it only improves the numbers before the fix.

Note
====

A lot of the time remaining comes from the aggregation of
draft/unpaid invoices with the correct rate done in python instead of in
SQL. This commit doesn't change the behavior but this could be rethought
from a function point of view.

________________________________________________________________________

[^1]: the old functions have been kept for compatibility, new ones are
suffixed by `_batched` and made private if it wasn't the case.
[^2]: some optimization require the views and indexes to be updated

closes odoo/odoo#103697

Related: odoo/enterprise#33165
Signed-off-by: Quentin De Paoli <qdp@odoo.com>