fix: constant folding for Nat.ble and Nat.blt

[TwoFX]

This PR fixes a bug in the constant folding for the Nat.ble and Nat.blt function in the old code generator, leading to a miscompilation.

Closes #6086

[TwoFX]

!bench

[leanprover-bot]

Here are the benchmark results for commit a2d37c6.
There were significant changes against commit a074bd9:

  Benchmark           Metric                    Change
  ================================================================
+ big_omega.lean MT   branch-misses              -1.7%   (-18.8 σ)
+ nat_repr            branches                   -1.3% (-1777.3 σ)
+ nat_repr            instructions               -1.9% (-2472.1 σ)
- rbmap               instructions                5.9%   (404.2 σ)
- rbmap_1             instructions                1.6%    (16.6 σ)
- rbmap_1             task-clock                  5.7%    (13.6 σ)
- rbmap_1             wall-clock                  5.7%    (13.6 σ)
- rbmap_10            instructions                4.5%    (93.7 σ)
- rbmap_fbip          task-clock                 12.5%    (11.8 σ)
- rbmap_fbip          wall-clock                 12.5%    (12.0 σ)
- rbmap_library       instructions                4.7%   (296.0 σ)
- stdlib              dsimp                       1.2%    (67.4 σ)
- stdlib              fix level params            1.1%    (32.2 σ)
- stdlib              process pre-definitions     1.1%    (18.8 σ)

[leanprover-community-bot]

Mathlib CI status (docs):

• ❗ Mathlib CI can not be attempted yet, as the nightly-testing-2024-11-15 tag does not exist there yet. We will retry when you push more commits. If you rebase your branch onto nightly-with-mathlib, Mathlib CI should run now. (2024-11-15 10:25:24)

[TwoFX]

!bench

[leanprover-bot]

Here are the benchmark results for commit e6ca7b5.
There were significant changes against commit a074bd9:

  Benchmark                   Metric          Change
  ===========================================================
- bv_decide_inequality.lean   branch-misses     1.2% (15.8 σ)
- bv_decide_mul               wall-clock        2.4% (11.6 σ)
- simp_arith1                 branch-misses     3.1% (20.3 σ)