Implement bitmask for SecondaryMap serialization

[mrowqa]

Small improvement I was supposed to implement earlier. Might help dtolnay/watt#2.

r? @sunfishcode

[bnjbvr]

Hey @mrowqa, can you give some context around this change: why is it useful, what does it change? (if it becomes lengthy, please open an issue!)

Is this covered by tests, and if not can you add some, please? Thanks!

[mrowqa]

The original reason was cache system in Wasmtime. It uses serde for serialization of compilation cache which consists of SecondaryMaps (among others). At the time of implementing the cache system, I had implemented simpler SecondaryMap serializer (current code) that uses whole byte for a bit flag (presence of an object at a given index) and this PR changes the serializer to pack these bit flags compactly. After implementing the simpler serialization, I had added compression to the cache system, so using whole bytes for single bit information shouldn't affect too much the size of the final cache that lands on user disk. I haven't measured the difference in performance and peak memory usage (intermediate step - after decompression, before deserialization) that this PR introduces.

The change is tested by Wasmtime cache system tests.

CC @sunfishcode

[alexcrichton]

Thanks for the PR again, and as a procedural note the Cranelift repository has now merged into the wasmtime repository.

PRs are no longer landing in this repository, and unfortunately there's no "one button" solution to move a PR to the wasmtime repository. A script has been prepared, however, to assist you in transferring this PR to the wasmtime repo. Feel free to reach out on Zulip with any questions!

[tschneidereit]

We'll archive this repository in favor of the Wasmtime repository — see Alex's comment about the merge above. I'm closing this PR now in order to have an explanation in a comment before archiving the repo auto-closes all PRs.