Benchmark existing ways to check for IDENTITY CIDs

[masih]

Benchmark existing ways to check for IDENTITY CIDs

Add benchmarks that compare two ways of checking for
multihash.IDENTITY code:

1. Cid.Prefix().MhType
2. Decode of Cid.Hash()

This benchmark illustrates that using Cid.Prefix is more efficient than
multihash.Decode. Users wishing to perform such a check should use
Cid.Prefix.

Consider that Cid.Prefix is already efficient enough and introducing a
dedicated API for performing this check will likely result in small
gains.

BenchmarkIdentityCheck
BenchmarkIdentityCheck/Prefix
BenchmarkIdentityCheck/Prefix-8         	351656472	         3.280 ns/op
BenchmarkIdentityCheck/MultihashDecode
BenchmarkIdentityCheck/MultihashDecode-8         	 5715622	       239.1 ns/op

Relates to #133

[mvdan]

You can drop the CidV1 part of the benchmark name; Cid is redundant since this module is go-cid, and v1 is redundant as we don't really care if the CID is v1 or not - it just so happens that all identity CIDs are currently v1. If they also supported a future CIDv2, then the benchmark would simply do both in each iteration.

You could also make the two methods sub-benchmarks, to simplify sharing code and make the output nicer. For example:

func BenchmarkIdentityCheck(b *testing.B) {
    // cid setup, ReportAllocs, etc
    b.Run("Prefix", func(b *testing.B) {
        b.RunParallel(func(pb *testing.PB) {
            // check via Prefix
        })
    })
    b.Run("MultihashDecode", func(b *testing.B) {
        b.RunParallel(func(pb *testing.PB) {
            // check via MultihashDecode
        })
    })
}

[mvdan]

SGTM! Probably wait for a review from a steward though.

[warpfork]

I have no reason to complain about more benchmarks :)

If you still have the context in your head: I wouldn't mind a short paragraph of comment attached to the function about what you were looking for / why it mattered / and what the rough findings were at the time you committed. (It'll get stale, sure -- but that doesn't mean it won't still be useful context clues to some future archaeologist^W reviewer.)

(... I guess that would be roughly the same content you've already included in the PR message. Still worth repeating in the diff, IMO. :) I find it a lot easier to rediscover links in a comment in the relevant code than... links that are in PRs that are maybe (depending on which merge button you click in github) mentioned by number in the commit log that you might find after you go blame-splunking.)

[masih]

Will do 🍻

Context captured here

TLDR; there are multiple ways of inferring if a CID is IDENTITY and one of them is more efficient. The one that's more efficient is efficient enough to avoid the need to implement new APIs to do this (see closed PR for more details)