[DPE-2651] Juju3 pipelines / Testing secrets

[juditnovak]

Issue

JujuVersion.has_secrets have to be correctly patched so that secrets-related code would be invoked

Solution

Pytest fixtures ensuring the usage to JujuVersion.has_secrets (instead of its default behavior returning False all way through)

NOTE: The 3 pipeline failures are unrelated to secrets. They will be addressed in upcoming PRs as below.
This PR will be merged AFTER those.

• self-healing test failures will be fixed in Avoid any actions when a unit is offline and allow mysql GR to recover #334
• upgrade test failures will be fixed after DPE-2654 profile-limits option #331

[codecov]

Codecov Report

Merging #333 (167f2a0) into main (1f72f1b) will increase coverage by 0.96%.
The diff coverage is 50.00%.

@@            Coverage Diff             @@
##             main     #333      +/-   ##
==========================================
+ Coverage   64.50%   65.46%   +0.96%     
==========================================
  Files          17       17              
  Lines        3113     3113              
  Branches      413      413              
==========================================
+ Hits         2008     2038      +30     
+ Misses        979      946      -33     
- Partials      126      129       +3     

Files	Coverage Δ
src/relations/mysql_provider.py	37.80% <50.00%> (ø)

... and 1 file with indirect coverage changes

[shayancanonical]

Looks amazing!

[carlcsaposs-canonical]

If we're planning to support multiple juju versions long-term, I think we might want to rework the fixture so that it's only for integration tests and unit tests for all juju versions run at the same time

But if we only need to support juju 2+3 for a short time this looks good

Also I think the skip vs deselect for tests might be a bit confusing with the UX that we're using to group integration tests, but since we're not heavily using groups we probably won't encounter that often for now

[paulomach]

excellent, thank you