fix: Move remove_if_else pass after second inlining

[vezenovm]

Description

Problem*

Resolves

Summary*

For the test added in this PR I bumped the max_iter in try_merge_only_changed_indices. We then get the following panic:

The application panicked (crashed).
Message:  internal error: entered unreachable code: All Value::Instructions should already be known during inlining after creating the original inlined instruction. Unknown value v41 = Instruction { instruction: Id(76), position: 0, typ: Array([Numeric(Unsigned { bit_size: 32 })], 30) }
Location: compiler/noirc_evaluator/src/ssa/opt/inlining.rs:253

We have to inline immediately following mem2reg before extra instructions are added by the remove_if_else pass.

Additional Context

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

Changes to circuit sizes

Generated at commit: 9e3d1aa53374f01e7d57db5ea2f2a4f6e13db3a4, compared to commit: a87c655c6c8c077c71e3372cc9181b7870348a3d

🧾 Summary (10% most significant diffs)

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
fold_numeric_generic_poseidon	-4 ✅	-28.57%	-4 ✅	-21.05%
fold_call_witness_condition	-4 ✅	-36.36%	-4 ✅	-23.53%

---

Full diff report 👇

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
regression_mem_op_predicate	56 (+12)	+27.27%	3,559 (+23)	+0.65%
conditional_1	6,040 (-1,131)	-15.77%	38,799 (0)	0.00%
no_predicates_numeric_generic_poseidon	55 (-4)	-6.78%	10,758 (-4)	-0.04%
slice_dynamic_index	1,353 (-158)	-10.46%	6,947 (-297)	-4.10%
slices	350 (-110)	-23.91%	3,420 (-198)	-5.47%
hashmap	176,062 (-1,608)	-0.91%	367,934 (-26,986)	-6.83%
fold_numeric_generic_poseidon	10 (-4)	-28.57%	15 (-4)	-21.05%
fold_call_witness_condition	7 (-4)	-36.36%	13 (-4)	-23.53%