Also, add logs in case we find a connection with a required 2FA so we can use it to
implement 2FA later on.

The message should help the user understand he didn't choose the proper
website (pro instead of par).

When the website is under maintenance, going to loginpage leads to a
404, so before that we go to main page to check the message.

Some connections still have the old layout so we need to handle both.

There have been several design changes on the accounts pages, so a lot
of xpaths had to be updated. The navigation seems unchanged.

The previous one now leads to a 302.
Arrival page is still the good one to parse accounts.

Previously we were getting the transactions of the deffered card on the comings of the main account

On CM, check_redirections() controls when 2FA is coming or not.
Website proposes to redo 2FA a few weeks
before its end of validity, also detected in check_redirections().
We can skip it, with skip_redo_twofa().
But for now it would only work in interactive mode.
Thankfully we have the timestamp of the end of the 2FA validity.
So now, we only need interactive mode if we are not in the validity
period, effectively using skip_redo_twofa() otherwise.

For the labels, we are fetching an <a> tag, except that some invests
have no link so we were retrieving invests with empty labels.
We now handle <a> or <strong> cases with a Coalesce.
Codes were handled with IsinCode & code_types with IsinType.
Used Base filter instead of ugly filter()(self)[0].xpath().

CraGr send bad JSON with patterns like '..."phoneNumber":",...'
Fix: updated build_doc()

CleanDecimal.French instead of CleanDecimal.SI is needed
for last_payment_amount attribute,
like every other amount attribute on the website.

Sometimes there are no transactions for 3 months, it's acceptable.
However 4 pages (1 year) without transaction is rarer.

Some accounts don't have an IBAN, which is the case for life insurances for example.

For some connections, the default authentification is by certificate and
the fallback method is EMV. So we can use the fallback request again to use
password authentification.

Not long after the last patch that set the timeout to 60s, the website
became even slower and it's not always enough.

When no par space is found, instead of retrying multiple times and raise
a WebsiteUnavailable, we just send an empty list of accounts

The iban is now directly accessible with the main account json, and so,
there is no need anymore to load the transfer page to get the iban.

Also, if fixes the issue of iban not being available for accounts that
were not allowed for transfers.

Filter for current month transactions
must also look at year since, upon changing year,
filtering only on month would fail.

Some documents are not ready to be downloaded, and when trying to download
them it results in a 500 error.
When now apply the same logic as in the site to decide if a document should
be listed or not.

Some documents may not be available yet, or may have been modified and are not
available.
Trying to download such documents results in a 500 error.
Thus we detect if the documents are available, using the same (javascript) logic
as the main site.