fix: Remove outdated barrier when automatically inserting drops

[kkoci]

Remove stale FuncDefn skip in insert_drops (#1516)

Removes a workaround in insert_drops (core.py) that skipped all FuncDefn nodes during drop insertion. The skip was
added to work around a num_out_ports bug in hugr (Quantinuum/hugr#2438), which has since been fixed in hugr 0.15.4 —
the version guppylang already depends on. The guard is also redundant because FuncDefn outputs are FunctionKind, never ValueKind, so no drop would have been inserted anyway.
Changes:

• Remove 4 lines from insert_drops in compiler/core.py (the stale comment referencing #2438, the isinstance(data.op, ops.FuncDefn) check, and the continue)
• Add regression test test_drop_with_multiple_funcdefns to tests/integration/test_drop_insertion.py

Closes #1516

[maximilianruesch]

Please remember that the issue title often does not suffice as a descriptive enough PR title, and they serve different purposes: An issue title often describes the problem or symptom, and the PR title should describe the solution / broader goal. This is in line with the knowledge about a task at the respective creation times.

I have amended your PR title for this one as an example.

[kkoci]

Thank You @maximilianruesch , so it is not about reproducing the issue title, rather a more descriptive (briefly) title about the approach/solution.

[maximilianruesch]

Yes. In essence, this title directly goes into the changelog, and it appears in the commit message in the git history (which is essentially a more detailed changelog). Whenever someone reads that, they want to have a summary of what has been changed or at least of what has been fixed (to a reasonable level of detail). "Closed issue mentioned in a comment" is neither descriptive nor helpful to someone reading the history, as it offers no explanation as to what has been done and why.

On the other hand, the current title offers several points: "Remove redundant" signals that something that is no longer needed was removed, and the change should have no impact on the semantics of the program; "when automatically inserting drops" signals the place of the change, thus making it easier to find for people looking for changes related to drop semantics.

[mark-koch]

Thanks! This looks good, however the CI checks are failing due to formatting. Can you please run ruff format and check that the CI passes?