fix: memleak caused by derived value analysis

[hargoniX]

No description provided.

[hargoniX]

!bench

[leanprover-radar]

Benchmark results for 67fb993 against fb1dc91 are in. Significant changes detected! @hargoniX

• 🟥 build//instructions: +17.1G (+0.14%)

Large changes (4🟥)

• 🟥 compiled/parser//instructions: +169.0M (+0.45%)
• 🟥 compiled/phashmap//instructions: +243.2M (+2.55%)
• 🟥 compiled/qsort//instructions: +2.0G (+10.68%)
• 🟥 elab/workspaceSymbols//instructions: +635.8M (+1.88%)

Medium changes (1✅, 3🟥)

• 🟥 compiled/workspaceSymbolsNewRanges//instructions: +6.2M (+0.80%)
• 🟥 elab/big_do//instructions: +88.4M (+0.37%)
• ✅ elab/big_omega//wall-clock: -125ms (-7.00%)
• 🟥 elab/cbv_arm_ldst//instructions: +934.2M (+1.24%)

Small changes (2✅, 33🟥)

Too many entries to display here. View the full report on radar instead.

[mathlib-lean-pr-testing]

Mathlib CI status (docs):

• ❗ Batteries/Mathlib CI will not be attempted unless your PR branches off the nightly-with-mathlib branch. Try git rebase fb1dc9112b2f6b66d54aaf39ac030dae7153c822 --onto 4bf7fa7447eea00cecba8327bb9c9e5f4485f0a7. You can force Mathlib CI using the force-mathlib-ci label. (2026-03-23 22:57:02)

[leanprover-bot]

Reference manual CI status:

• ❗ Reference manual CI will not be attempted unless your PR branches off the nightly-with-manual branch. Try git rebase fb1dc9112b2f6b66d54aaf39ac030dae7153c822 --onto cfa8c5a036d6990635c6ec50b02d0e806995cec3. You can force reference manual CI using the force-manual-ci label. (2026-03-23 22:57:04)