fix: Update broadcast exchange logic to support reused exchanges

[andygrove]

Which issue does this PR close?

N/A

Rationale for this change

This fix was needed to fix some Spark SQL test failures in #1736.

I am extracting this to a separate PRs to cut down on the review burden in #1736.

What changes are included in this PR?

Two one-line fixes to recurse into reused exchanges when looking for broadcast exchanges.

How are these changes tested?

We do not have any tests in Comet that would expose this bug, and the relevant Spark SQL tests currently fall back to Spark (until #1736 is merged).

I did spend time trying to write a repro but have not been successful so far.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0% with 2 lines in your changes missing coverage. Please review.

Project coverage is 59.39%. Comparing base (f09f8af) to head (cdff050).
Report is 246 commits behind head on main.

Files with missing lines	Patch %	Lines
...he/comet/rules/EliminateRedundantTransitions.scala	0.00%	0 Missing and 1 partial ⚠️
...pache/spark/sql/comet/CometColumnarToRowExec.scala	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #1858      +/-   ##
============================================
+ Coverage     56.12%   59.39%   +3.26%     
- Complexity      976     1151     +175     
============================================
  Files           119      130      +11     
  Lines         11743    12669     +926     
  Branches       2251     2378     +127     
============================================
+ Hits           6591     7525     +934     
+ Misses         4012     3930      -82     
- Partials       1140     1214      +74     

☔ 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.

[andygrove]

This PR is no longer needed now that the diff in #1736 is much smaller