Do not promote mean and covariance of `MvNormal` to common element type

[devmotion]

Ref #1964. Currently just to see if and what it would break.

[hersle]

Thanks for giving it a go!

How bad is this? I can imagine several of these test failures are fixable.

What really matters at the end of the day? That the core API functions are type-stable and respect eltype?

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 86.24%. Comparing base (f33cc10) to head (12241fe).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1965      +/-   ##
==========================================
- Coverage   86.26%   86.24%   -0.02%     
==========================================
  Files         146      146              
  Lines        8763     8762       -1     
==========================================
- Hits         7559     7557       -2     
- Misses       1204     1205       +1     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

:rocket: New features to boost your workflow:

• :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[hersle]

Is this good to merge, or are there more things to consider?

[devmotion]

No, this is a major change, so it requires a bit more discussion, feedback from other maintainers, and should in the end be approved by another maintainer.

I'm also not sure yet whether the workaround for UniformScaling is desirable; maybe that codepath should be deprecated completely in favour of FillArrays.Eye which in contrast to UniformScaling has a clearly defined size.