Update loss checks

[alyst]

The PR slightly tweaks the recently introduced checks that the SEM loss function uses the compatible SemObserved object:

• Instead of throwing an error, it allows using SemObservedMissing for SemML and SemWLS, but shows a warning that approximate cov/mean will be used.
  In my use cases it is actually unfeasible to use FIML, because of the large number of observed vars and missing patterns, but EM-based approximation works quite well with SemML
• The missingness check in SemObservedData is tweaked to avoid generating temporary array, which could be quite large
• The messages introduced by these checks are multiline and contain indentation + they respect the indentation of the surrounding code.
  The correct way to define such messages is to use indented """ strings -- the indentation of """ is subtracted from the string indentation, so they are not overindented when printed.
• I have also switched to the pipe syntax for these messages, because it results in fewer indentations + potentially more readable overall

[codecov]

Codecov Report

❌ Patch coverage is 33.33333% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 71.68%. Comparing base (1f8d2a9) to head (03b2048).
⚠️ Report is 30 commits behind head on devel.

Files with missing lines	Patch %	Lines
src/loss/WLS/WLS.jl	33.33%	2 Missing ⚠️
src/loss/ML/ML.jl	0.00%	1 Missing ⚠️
src/observed/data.jl	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            devel     #309      +/-   ##
==========================================
- Coverage   71.83%   71.68%   -0.16%     
==========================================
  Files          51       51              
  Lines        2223     2239      +16     
==========================================
+ Hits         1597     1605       +8     
- Misses        626      634       +8     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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

[Maximilian-Stefan-Ernst]

Thank you! Also good to know about the indentation behaviour.