chore: harden markPersistent uses

[Kha]

This API may or may not have been a footgun, better to be safe than sorry

[Kha]

!bench

[leanprover-bot]

Here are the benchmark results for commit 832f568.
The entire run failed.
Found no significant differences.

[Kha]

!bench

[leanprover-bot]

Here are the benchmark results for commit 3cd5f18.
There were significant changes against commit 5f1ff42:

  Benchmark                  Metric                    Change
  =====================================================================
- big_do                     branches                    1.8%  (71.9 σ)
- big_do                     instructions                1.7%  (72.6 σ)
+ big_omega.lean             branch-misses              -3.9% (-14.8 σ)
- big_omega.lean             branches                    1.6%  (39.0 σ)
- big_omega.lean             instructions                1.4%  (48.7 σ)
+ bv_decide_mul              branch-misses              -2.2% (-10.6 σ)
- lake env                   task-clock                  7.7%  (30.6 σ)
- lake env                   wall-clock                  7.6%  (37.8 σ)
- parser                     task-clock                  5.2%  (35.1 σ)
- parser                     wall-clock                  5.2%  (35.6 σ)
- stdlib                     attribute application       2.6%  (18.5 σ)
- stdlib                     instructions                1.2% (512.5 σ)
- stdlib                     process pre-definitions     2.3%  (10.7 σ)
- tests/bench/ interpreted   instructions                1.9% (883.4 σ)
- tests/bench/ interpreted   task-clock                 30.2%  (26.3 σ)
- tests/bench/ interpreted   wall-clock                 13.3%  (22.8 σ)

[leanprover-community-bot]

Mathlib CI status (docs):

• ❗ Batteries/Mathlib CI will not be attempted unless your PR branches off the nightly-with-mathlib branch. Try git rebase 5f1ff42a15eec6fb7696435f86bae82500bd3de5 --onto 6d495586a1836da3e12efb4fbf9946bd9088ac5f. (2024-11-29 14:36:10)