Inline symbol-related methods

[Amanieu]

Together with m4b/scroll_derive#6, this significantly improves the performance of SymIterator.

When iterating through an ELF file with 2.5M symbols:
Before: 219ms
After: 120ms

[philipc]

How did you decide which functions to inline? Inlining functions is fine, just trying to gauge the completeness of this, or whether maybe some of these functions are big enough that maybe they shouldn't be inlined.

[m4b]

Yea I don't think this many functions need to be inlined to speed up symbol iteration (probably just the try_from is my guess, taken care of in the scroll PR), e.g., is_import, is_function, size_with, from_raw, etc., shouldn't be need to be inlined

[Amanieu]

I just marked functions which had a very simple implementation (e.g. is_function) or that can significantly benefit from constant propagation or loop unswitching (e.g. try_from_ctx where the match on Container can be moved out of the hot loop by the compiler).

[philipc]

This is the only one that I don't see any benefit for, since it won't be called often and there's little room for optimization.

[Amanieu]

This method only performs 3 copies and 2 zero-initializations, which are practically free if inlined into the parent function. While you are technically correct that this won't be called often, there is absolutely zero downside to marking it as inline.

[philipc]

Thanks!