Add type checking to graph construction

[beojan]

Add type checking to products (at least when registering algorithms in C++).

The qualified_name for output products, and the specified_label for input products now have type information attached (in the form of a custom type_id currently calculated mostly at compile time) that is compared in find_products.

I had to use a custom type_id so we could do things like treating handle<int>, int const&, and int const* as the same type. This should also simplify extending this type checking to Python algorithms.

I think I've done the best I can with code coverage. I don't know why it thinks the formatting code isn't covered, it definitely should be in my type_id.cpp test.

[codecov]

Codecov Report

❌ Patch coverage is 85.99034% with 29 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
phlex/core/edge_creation_policy.cpp	81.81%	6 Missing ⚠️
phlex/model/type_id.hpp	92.77%	2 Missing and 4 partials ⚠️
phlex/core/declared_transform.cpp	0.00%	2 Missing and 1 partial ⚠️
phlex/core/declared_unfold.cpp	0.00%	3 Missing ⚠️
phlex/core/product_query.cpp	80.00%	3 Missing ⚠️
phlex/core/product_query.hpp	75.00%	2 Missing and 1 partial ⚠️
phlex/model/product_specification.cpp	88.46%	2 Missing and 1 partial ⚠️
phlex/model/product_specification.hpp	75.00%	2 Missing ⚠️

@@            Coverage Diff             @@
##             main     #119      +/-   ##
==========================================
+ Coverage   80.31%   81.29%   +0.97%     
==========================================
  Files         115      116       +1     
  Lines        1910     2042     +132     
  Branches      302      326      +24     
==========================================
+ Hits         1534     1660     +126     
  Misses        247      247              
- Partials      129      135       +6     

Flag	Coverage Δ
unittests	81.29% <85.99%> (+0.97%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
phlex/core/declared_fold.cpp	100.00% <ø> (ø)
phlex/core/declared_fold.hpp	92.98% <100.00%> (ø)
phlex/core/declared_observer.cpp	33.33% <ø> (ø)
phlex/core/declared_observer.hpp	100.00% <ø> (ø)
phlex/core/declared_predicate.cpp	50.00% <ø> (ø)
phlex/core/declared_predicate.hpp	89.65% <ø> (ø)
phlex/core/declared_transform.hpp	91.48% <100.00%> (+0.18%)	⬆️
phlex/core/declared_unfold.hpp	94.54% <100.00%> (+0.20%)	⬆️
phlex/core/detail/filter_impl.cpp	92.50% <ø> (ø)
phlex/core/detail/filter_impl.hpp	100.00% <ø> (ø)
... and 21 more

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4e67f8c...434c826. Read the comment docs.

🚀 New features to boost your workflow:

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

[knoepfel]

@beojan, bravo. I think you have hit all of the spots in the implementation that would be affected by imposing type constraints when establishing edges. I have several comments below for your consideration and for further discussion. A great start—thanks.

[beojan]

I ran the coverage script locally using the LLVM backend and that shows line coverage for type_id.hpp at 96.9%

[beojan]

OK, should be ready now. I have some ideas for further cleaning up find_producer but that's for another PR.

[knoepfel]

@beojan, apologies for the delay in reviewing this. I think the highest priority item is to implement a unit test verifying that nodes with mismatched types are not connected (see below). Happy to discuss any of my comments.