Handle Symmetric/Hermitian in PDSparseMat

[timholy]

Packages like NLPModels return Hessians as Symmetric(L, :L) where L
is a sparse lower-triangular matrix. Previously, PDSparseMat only
supported SparseMatrixCSC directly.

This also:

• Acknowledges that CholTypeSparse is now an abstract type and adds
  its concrete type as a parameter to PDSparseMat.
• Adds a "coercion cascade" PDSparseMat{T}(args...) -> PDSparseMat{T, S}(args...) -> PDSparseMat{T, S, C}(args...) to force type conversion. This is sometimes needed when one object of a different
  type needs to be converted into some standard type in generic code.
• Moves size checks into the inner constructor, so that it is impossible
  to create an invalid PDSparseMat.

I expect this to pass tests on 1.10 but fail on 1.12 due to JuliaSparse/SparseArrays.jl#662 (UPDATE: fix already merged, queued for backporting). Before going further, it makes some sense to take a step back and acknowledge other caveats:

• I had to disable one inferrability test because some internal Julia code is inferred to return a Union
• a simpler (but perhaps less computationally efficient) approach is to create a new, symmetric SparseMatrixCSC from the one that was supplied.
• finishing Unify PDMat and PDSparseMat + move SparseArrays support to an extension #188 might be a better way forward

[codecov-commenter]

Codecov Report

❌ Patch coverage is 94.44444% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 92.60%. Comparing base (61c65b1) to head (a9f3916).

Files with missing lines	Patch %	Lines
src/pdsparsemat.jl	94.11%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #230      +/-   ##
==========================================
- Coverage   93.37%   92.60%   -0.78%     
==========================================
  Files          11       11              
  Lines         800      811      +11     
==========================================
+ Hits          747      751       +4     
- Misses         53       60       +7     

☔ 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.