Breaking changes for v4 #351

I think BioSequences has accumulated enough warts and cruft since its v3 release in 2022 that it warrants a new breaking release.
I do recognize that breaking changes are disruptive. However, given that migration can now be automated with an AI agent given a good migration guide, I think it's more important to get the semantics right.

This is a long term project. I do not expect this to land in 2026, and maybe not even in 2027. It's possible this will not land until Julia 1.15 (at which point it is likely Julia 1.14 has been declared LTS).

• Switch LongSequence and LongSubSeq to SeqBuf, SeqView and LongSeq as discussed in Discussion: Make LongSequence fixed-length? #329. Maybe retain type aliases LongSequence and LongSubSeq. Back all these types by RefVector from MemoryViews.jl
• Review geneticcode.jl. Drop the AbstractDict subtype of GeneticCode. Maybe expose the creator of genetic codes as public. We should not have GeneticCode <: AbstractDict{UInt64, AminoAcid} #347
• Redesign the AlphabetCode system and Kmer's RecodingScheme into one coherent trait-based copying system. We probably want one dispatch mechanism (RecodingScheme) between source type and destination alphabet. This should also encompass a mechanism for opting into ASCII encoded alphabet types. This is related to Document BioSequence interface(s) #328
• Review functions taking NucleicAcidAlphabet which makes assumptions that actually only hold for DNAAlphabet and RNAAlphabet, and restrict these signatures
• Remove convert from BioSequence (Remove convert for biosequences #321)
• Hide inner constructors (Hide inner constructors of objects in this package #318)
• Degrade Random to a weakdep (Consider making Random a weak dep #289)
• Resolve Design of hash(::BioSequence). What to do? #243 by making generic BioSequence not comparable and re-implementing hashing. Maybe look at AbstractString's hashing in Base which is fast. Make the SeqBuf / LongSeq / SeqView comparable.
• Review majorityvote and seqmatrix - do we want them?
• Consider a thorough aliasing policy for BioSequences. How do we handle aliasing arguments?
• Expose this package to a thorough API review by an AI agent for suboptimal APIs that are candidates for change