chore(licm): Break things up further in LICM

[aakoshh]

Description

Problem*

LoopContext had a bunch of current_block_* fields which had to be reset for each block we iterated over in hoist_loop_invariants, and assigned new values in various methods that also mutated other fields on LoopContext. Some were unrelated, like is_control_dependent_post_pre_header also set the block purity, seemingly just because all_predecessors was available there. This made it hard to find where various flags on LoopContext came from.

Summary*

• Breaks out the current_block_* fields from LoopContext into a BlockContext and introduces an LoopInvariantContext::init_block_context method to create it, so we cannot forget to set the values, and there is no risk of reusing something from the previous iteration.
• Changes all query methods for the BlockContext initialisation to avoid mutating the LoopContext, so it's a bit more obvious where things are set, all clear in the init_block_context method.
• Renames get_values_defined_in_loop to init_loop_context as it did more than what it said on the tin.
• Makes LoopContext::pre_header non-optional.
• Copied some extra explanation for control dependence from the paper to have an intuitive definition.
• Avoid checking repeatedly whether the bounds in the LoopContext guarantee execution by storing the result in a field.

Additional Context

I haven't confirmed, but the fact that is_control_dependent_post_pre_header exited early if the block was part of a known control dependent nested loop, and then it did not set the is_current_block_impure flag, was possibly a bug. Now that they are separate, the impurity flag is set in all cases.

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]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Test Suite Duration'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: d09fa36	Previous: a38288a	Ratio
test_report_zkpassport_noir_rsa_	2 s	1 s	2

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[vezenovm]

Overall looks good! Just have that one request for removing the ControlDependenceResult variant with the lifetime.