Simplify the intersection #415 asks for, and correct two tests from #666 (#415, #424, #263)

[Rafael-SOWNet]

Following up on your review of #666, where three of the tests I added were pinning
something other than what the issue was about. Thank you for catching them — you were
right in each case.

#415 — the screenshotted example

The screenshot is

(-1; 1) /\ ((-(sqrt(33) + 3) / 6; (sqrt(33) - 3) / 6) \/ (1; +oo))

and it came back exactly as written. Two things were missing, and both are here:

• An intersection did not distribute over a union. Nothing could be done with the
  right-hand side however simple each piece was.
• IntersectIntervalAndInterval gave up unless every endpoint was a bare Real node.
  (sqrt(33) - 3) / 6 is a Divf, so it never was one, and the whole intersection was
  abandoned. Endpoints are now compared by what they evaluate to, while the bounds of
  the answer are taken from the original expressions — so the result stays exact
  rather than becoming a hundred decimal places.

→ (-1; (sqrt(33) - 3) / 6)

The test is now that example, rather than an intersection I chose myself.

#424 — it was an incorrect test, as you said

I had tested "5 provided x = 3", which is a Providedf. The issue is about the
Piecewise node, which is a different one — so the test was not testing the issue at
all. Rewritten with piecewise(...), including a three-branch case that checks the
branches are taken in order.

It passes. The feature does work; my test simply never exercised it.

#263 — the limit is now asserted, not just its termination

It only checked that the limit came back. The limit is what that expression was written
for, so its value is asserted: 167403915.

Still outstanding from that review

• Equation system solver fails on simple case #550 — you said the null-returning line is the one to fix. I have not touched it
  here: SolveSystem returns null when it finds nothing, and what it should return
  instead — an empty matrix, an empty set, something else — is an API decision I would
  rather you made than guess at. Try another equation when elimination stalls, fixing dense linear systems (#608) #667 fixed one reason it was reaching that line
  needlessly, but the line itself is untouched. Happy to do it once you say which.
• (x+1)^2(x+2-1)^2 cannot be expanded #164 is fixed in Add up the like terms Expand produces (#164) #676, separately, since it turned out to be a real defect in
  Expand rather than a test problem.

Testing

UnitTests 3886 passed / 0 failed, FSharpWrapperUnitTests 127 passed / 0 failed, both
netstandard2.0 and net7.0 build.

🤖 Generated with Claude Code

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 88.88889% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.45%. Comparing base (90c00a8) to head (de0d8ee).
⚠️ Report is 57 commits behind head on master.

Files with missing lines	Patch %	Lines
...Core/Entity/Omni/Sets/SetOperators.Intersection.cs	90.00%	0 Missing and 1 partial ⚠️
...Functions/Simplification/Patterns/Patterns.Sets.cs	87.50%	1 Missing ⚠️
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #677      +/-   ##
==========================================
- Coverage   80.99%   80.45%   -0.55%     
==========================================
  Files         155      156       +1     
  Lines       13687    12925     -762     
  Branches     1957     2122     +165     
==========================================
- Hits        11086    10399     -687     
+ Misses       1990     1919      -71     
+ Partials      611      607       -4     

☔ View full report in Codecov by Harness.
📢 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.