Simpler module structure

[chshersh]

After working with the tomland code for a while, it was discovered that its codebase is not optimal, and sometimes confusing to navigate. Its is a proposal to change the existing module structure. This is a breaking change, but since we recommend to import the top-level module qualified, almost all users of tomland shouldn't even notice the change.

My current proposal is the following:

• Rename: Toml.Bi.Combinators to Toml.Codec
  • This module contains all codecs , so it name could help find them
• Rename: Toml.Bi.Monad to Toml.Monad
• Rename: Toml.Bi.Map to Toml.BiMap
• Rename: Toml.Bi.Code to Toml.Decoding
  • It has both functions encode and decode, so Decoding is not precise enough... Maybe we should have two modules: Toml.Encode and Toml.Decode? Or maybe we can even combine Encode with Toml.Printer?
• Delete: Toml.Bi
  • Module structure is becoming flatter
• New: Toml.Key
  • This data is inside the Toml.PrefixTree module, and keep forgetting about that. Probably we can move it out to help navigation

Open questions:

1. The core polymorphic monad is called Codec. Wouldn't it be confusing to have Toml.Monad have the Codec type and the Toml.Codec module to contain all combinators? Maybe names should be Toml.Codec and Toml.Combinators instead?
2. Maybe we can even remove the Toml.Type prefix and remove the .Type part from all Toml.Type.* modules?
3. What to do with Toml.Bi.Code? 🙂

Any ideas of other improvements are appreciated 🙂

[chshersh]

New structure after discussion with @vrom911:

• Have 3 top-level namespaces: Toml.Codec, Toml.Parser and Toml.Type
• Everything else will go in those namespaces