Rework letters into submodules, add pluggable funcs and parser options

[rorycl]

This is a rewrite of letters to use submodules with local testing and provide parser options together with pluggable funcs.

The main benefits of the rewrite is to allow:

• the use of methods on Parser to share state while processing
• support for pluggable funcs, most notably to allow custom file filtering and/or parsing to avoid reading whole files into memory.
• the provision of options to customise the operation of the parser.
  See opts_test.go and the optpkg_test package tests for usage examples.
• a more modular approach in general which will hopefully encourage incremental improvements to be made in future.
• a reduction in code and addition of comments.

The use of string types for email Content* information is, I think, still a bit of a mess and could be further rationalised since very little of this information is exposed to the user. Where it is exposed it would be better to simply use strings and simple string maps for easier usage.

I'm not sure we've been able to have a fruitful discussion about the above topics. I hope nevertheless there are some useful ideas here to help make the most of the very clear benefits of letters. It seems more fun and reasonable to collaborate, but in the absence of a chance for a discussion I'll probably make a hard fork.

To try this PR out you may need to use go workspaces since the submodules ("email", "decoders" and "parser") don't exist in the published letters namespace.

[rorycl]

Hi @mnako

With this addditional patch set I've now abstracted the Content-* capabilities into a new struct and methods.

Email headers and files now report their content information in structs along the following lines:

ContentInfo: &email.ContentInfo{
    Type: "text/plain",
    TypeParams: map[string]string{
        "charset": "ascii",
    },
    Disposition:       "",
    DispositionParams: map[string]string(nil),
    TransferEncoding:  "quoted-printable",
    ID:                "",
    Charset:           "ascii",
},

I have regenerated all the tests using utils/testRegen.go.

To try out this branch you may need a go.work file along the following lines:

replace github.com/mnako/letters/parser@latest => ./parser
replace github.com/mnako/letters/email@latest => ./email
replace github.com/mnako/letters/decoders@latest => ./decoders

use (
    .
    ./email
    ./parser
    ./decoders
)

[rorycl]

I've made my fork. I hope you find a few ideas that are useful in my code reorganisation. Thanks for the great work on decoding in particular!