Bitcode serde in benchmark

[CarsonBurke]

bitcode with default encode/decode is very different from bitcode serde in terms of performance. Including bitcode with feature = "serde" in the benchmark would be nice to be able to get a clear picture of the comparison to other serde serializers.

[finnbear]

Roughly speaking, serde-bitcode operates at the speed of binary serde formats with the size of bitcode-derive.

IMO it's fine to include in benchmarks in a row clearly labeled like this: "serde bitcode x.y.z"

Or like this (all numbers made up, for example purposes):

+---------+--------+-------+-----+
| name    | ser    | de    | ... |
+---------+--------+-------+-----+
| bitcode |  100%  | 100%  | ... |
|         |   31%* |  27%* | ... |
+---------+--------+-------+-----+

Where hovering the * numbers would show "serde version" (kind of like rkyv validated vs. unvalidated)

[CarsonBurke]

I think that would be a nice addition. Thank you for the further explanation.