[WIP] [experimental] replace ring gogoproto with csproto

[francoposa]

What this PR does:

Which issue(s) this PR fixes:

Fixes #

Checklist

• Tests updated
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[francoposa]

A quick run of benchmarking the Ring Clone since we didn't have a benchmark for it:

func BenchmarkRing_Clone(b *testing.B) {
	for _, numInstances := range []int{50} {
		for _, numZones := range []int{1, 3} {
			for _, shardSize := range []int{0, 3, 10, 30} {
				b.Run(fmt.Sprintf("num instances = %d, num zones = %d, shard size = %d", numInstances, numZones, shardSize), func(b *testing.B) {
					benchmarkClone(b, numInstances, numZones, 128, shardSize, false)
				})
			}
		}
	}
}

func benchmarkClone(b *testing.B, numInstances, numZones, numTokens, shardSize int, cache bool) {
	// Initialise the ring.
	ringDesc := &Desc{Ingesters: generateRingInstances(initTokenGenerator(b), numInstances, numZones, numTokens)}

	b.ResetTimer()

	for n := 0; n < b.N; n++ {
		ringDesc.Clone()
	}
}

output-ring-test-bench-ring-clone-edition2023-implicit.txt
output-ring-test-bench-ring-clone-main.txt
ring-clone-benchstat.txt

[francoposa]

Benchmark stats for all benchmarks in ring/ring_test.go - no particular effort put into optimization.

I also added a branch for vanilla proto edition 2023 using implicit presence to be compatible with proto3.
The generated code is different than doing proto3 codegen + csproto fastmarshal, but it doesn't make much difference performance-wise vs the proto3 + csproto version because none of these benchmarks cover marshaling/unmarshaling

output-ring-test-bench-ring-test-all-csproto-locked.txt
output-ring-test-bench-ring-test-all-edition2023-implicit.txt
output-ring-test-bench-ring-test-all-main.txt
ring-test-all-benchstat.txt