refactor: wrap shared memory buffer

[DaniPopes]

This wraps the shared memory Rc<RefCell<Vec>> in a SharedMemoryBuffer type so the Rc/RefCell details are kept in one place. LocalContext and SharedMemory now use the wrapper directly, while callers borrow initialized memory through slice-returning APIs.

This is only a cleanup around ownership and borrowing; it keeps normal Vec resizing behavior and avoids changing the buffer allocation/zeroing strategy.

Verified with cargo fmt --all, cargo test -p revm-context-interface --lib, cargo test -p revm-interpreter call_inputs --lib, and cargo check -p revm-context-interface -p revm-context -p revm-interpreter --all-features.

[codspeed-hq]

Merging this PR will improve performance by 3.63%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
✅ 175 untouched benchmarks
⏩ 1 skipped benchmark¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	MSTORE_50	20.1 µs	19.4 µs	+3.63%

---

_{Comparing DaniPopes:zeroed-shared-memory-buffer (853a17b) with main (1085ccf)}

Footnotes

1. 1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩