PURPOSE
Various improvements of the results' statistics were required, mainly regarding
* the filtering features that were not clear and impractical
* the way answer scores was reported
* the vertical space taken by questions never answered

Among other smaller changes, the layout of `datetime`-answers questions is
harmonized with the other answer-types (grouped by value), considering that
the "time" part is important and that the "seconds" part bof that time would
only be changed if it is relevant to the question.

SPECIFICATIONS
**Filters**
It is now more convenient to filter surveys based on:
- Whether they are fully completed
- Whether they are passed
- Whether they are failed (last 2 are mutually exclusive, and only available
for scored tests)

A badge is added to show the number of surveys that match each filter option
if chosen. Note that this simple implementation does not allow to show the
number of surveys which would be displayed if the filter was removed, or
replaced, so the badges are not shown once a filter is selected.

The question/answer filters can now be removed individually.
The i.fa-filter elements used to activate them are now always visible, and are
only available for answers that have been selected at least once.

**Scoring**
Scores are added in tables next to the answers instead of as additional lines
below the question caption.

Right answers are now marked in the Graph and Data tabs instead of being simply
listed under the question.

**Unanswered questions**
To make this special case more clear to the user, a discrete alert is added
below the question. The Graph and Table tabs are removed in this case, as they
contain no information. For Multiple choice and Matrix questions, the Table tab
stays visible to show the possible answers available.

*Misc*
To follow common representation, numbers which are equal to integer values will
show as integers in answers and scores (e.g., 30.0 will show as 30), using a
sub-template.

Distinguishing the questions and results from the previous/next ones was
difficult, hence the added margins and discrete lines.

Similarly to other fields, datetime results now display one line per answer
value, with its occurrence.

Stylesheets are cleaned up.

LINKS
Task-2618399
See odoo/upgrade#3168

The current SLA status analysis report is hard to understand and doesn't provide much added value.
As such, this commit focuses on providing more relevant information to the user.

This commit will:

 - Remove the 'SLA Status Failed' measure
 - Rename 'ticket failed' into 'Failed Tickets'. This measure will represent the number of tickets
   with at least one SLA failed
 - Add a 'Success Rate' measure that will represent the % of tickets closed without failing the SLA
   vs. the total number of tickets
 - Add a 'Failure Rate' measure that will represent the % of tickets closed and having failed the SLA
   vs. the total number of tickets
 - Add a 'Successful Tickets' measure that will represent the number of tickets closed without failing the SLA
 - Add a 'Tickets in Progress' measure that will represent the number of tickets with this SLA in progress
 - Change the 'day to reach SLA' measure to 'Hours to reach SLA'
 - Remove the 'open time (hours)' measure
 - Rename 'time to first assignment (hours)' into 'Hours to Assign'
 - Rename the 'Time to close (hours)' measure into 'Hours to Close'
 - Format the values from the measures expressing hours as 00:00 instead of 0
 - By default:
     - The filter will be 'last 30 days'
     - The report will be grouped by Helpdesk Team > SLA
     - The measures will be: 'tickets failed' + 'failure rate'

task-2500895