Cache field offset calculations using CIL

[sim642]

Closes #1964.

TODO

• Actually add caching.
• Benchmark.

[Copilot]

Pull request overview

This PR introduces memoization for CIL field offset computations to avoid repeated expensive Cil.bitsOffset calls, addressing #1964.

Changes:

• Updated GobHashtbl.find_or_add_default_delayed to pass the key to the default thunk.
• Added cached Cilfacade.fieldBitsOffsetOnly / fieldBytesOffsetOnly helpers and ensured the cache is cleared in Cilfacade.reset_lazy.
• Switched multiple call sites to use the cached field offset helpers instead of recomputing offsets.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/util/std/gobHashtbl.ml	Changes find_or_add_default_delayed to provide the lookup key to the default function.
src/framework/xsltResultOutput.ml	Adapts find_or_add_default_delayed call sites to the new default signature.
src/common/util/cilfacade.ml	Adds field offset caching and resets it via reset_lazy.
src/cdomains/unionFind.ml	Uses cached field bit offsets when mapping bit offsets back to struct fields.
src/cdomain/value/cdomains/offset.ml	Uses cached field bit offsets in offset-to-index computation (incl. bitfield handling).
src/analyses/memOutOfBounds.ml	Uses cached field byte offsets for field offset conversion.
src/analyses/base.ml	Uses cached field byte offsets for container_of-style offset cancellation logic.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[sim642]

This speeds up rsync analysis by ~20% (on top of the CIL hasAttribute fix): goblint/bench@6e4ada7.
So I'll merge this and we'll see from knightly if any of it carries over to sv-benchmarks.

[sim642]

we'll see from knightly if any of it carries over to sv-benchmarks.

There's no impact on sv-benchmarks: #tum-nightly-svcomp > commit 82ff50a.