optimize encoder for common types

[achille-roussel]

This leads a small throughput increase on the encoder benchmark and enables a memory optimization on the application side:

benchmark                  old ns/op     new ns/op     delta
BenchmarkCodeEncoder-4     12958944      12751068      -1.60%

benchmark                  old MB/s     new MB/s     speedup
BenchmarkCodeEncoder-4     149.74       152.18       1.02x

benchmark                  old allocs     new allocs     delta
BenchmarkCodeEncoder-4     0              0              +0.00%

benchmark                  old bytes     new bytes     delta
BenchmarkCodeEncoder-4     0             0             +0.00%

[f2prateek]

Breaking change? If so, should we leave them in for compatibility and add a deprecation note?

[achille-roussel]

Indeed, these methods were here for optimization purposes but the new form is more generic. I know that we have no software that depended on those methods, there may be something out there that could be depended on it, in that case they'll need to adapt their code.

I'll add something to the readme explaining that change that needs to be made, but I don't think a deprecation period would be overkill for this case.

[achille-roussel]

@f2prateek are you satisfied with the README update?

[f2prateek]

@achille-roussel yep!

[achille-roussel]

Thanks for the review!