Introduce "force_case_insensitivity" option and deprecate "case_sensitive" in StringFilter

[phansys]

Subject

Introduce "force_case_insensitivity" option and deprecate "case_sensitive" in StringFilter.

I am targeting this branch, because these changes respect BC.

Closes #1415.
Reverts #1395.

Changelog

### Added
- "force_case_insensitivity" option to `StringFilter` in order to force the database to ignore the case sensitivity when matching filters.

### Deprecated
- "case_sensitive" option in `StringFilter`.

[VincentLanglet]

Why null ?
What's the difference between null and false ?

If I understand the option, the only allowed value would be false and true

[phansys]

We haven't other way to detect if this option is set or not. Currently, I'm using the value from "case_sensitive" if this option is not set.
I think we have 2 choices: add a hasOption() method, or leave this value adding a NEXT_MAJOR comment.
WDYT?

[VincentLanglet]

Why do you need to know if the option is set ?

If case_sensitive is not null, use this (deprecated) value
If case_sensitive is null, use force_case_insensitivity, which is by default to false.

[phansys]

Because I'd like the new option to be respected if it is set, regardless the value of the deprecated option.

[VincentLanglet]

Then we could add a NEXT_MAJOR comment to change the default value to false and remove the true === or true !== checks. But It feel weird to provide a null value and a false value which has the same behavior. And introducing a null default value which is already deprecated and will be changed to false in the next major...

But when I read force_case_insensitivity => false, case_sensitive => false:
I would expect

• force_case_insensitivity won't force the case insensitivity so it will do nothing. (It never was said that false will force the case sensitivity)
• case_sensitive will force the case insensitivity because it's already the behavior of this option

[phansys]

But when I read force_case_insensitivity => false, case_sensitive => false:
I would expect

force_case_insensitivity won't force the case insensitivity so it will do nothing. (It never was said that false will force the case sensitivity)
case_sensitive will force the case insensitivity because it's already the behavior of this option

I think the idea transmitted by the name case_sensitive leads to wrong assumptions. In fact, that's the reason for this change.
That's why I feel like a more natural behavior to completely ignore the option if its counterpart (force_case_insensitivity) is set.

[phansys]

The NEXT_MAJOR comment about the default value was added.

[phansys]

@VincentLanglet, I've added a comment in the upgrade document to make this behavior explicit.