Adds decoder functionality

[jacobparry]

Summary

• Adds UXID decoder functionality for parsing existing UXIDs
• Provides separate UXID.EctoType module with proper typing for typed_ecto_schema compatibility https://github.com/bamorim/typed_ecto_schema
• Maintains backward compatibility with existing encoder functionality

Changes Made

• New decoder module (lib/uxid/decoder.ex) - supports both uppercase and lowercase UXIDs
• Enhanced main module - adds UXID.decode/1 function while preserving existing API
• Case parameter support - both modules now support the new case option from v2.0+
• Comprehensive test coverage - 46 tests with full coverage for decoder and EctoType functionality

Implementation Details

Case Support

• Decoder handles both uppercase and lowercase UXIDs seamlessly
• Both Ecto implementations support the new case parameter
• Mixed case UXIDs (edge cases) are handled correctly

Test Coverage

• ✅ 46 tests total (8 new tests added)
• ✅ All existing functionality preserved
• ✅ New decoder tests for both uppercase and lowercase
• ✅ Mixed case decoding edge cases
• ✅ Comprehensive stress tests

Usage Examples

# Standard usage (existing)
defmodule MyApp.User do
  use Ecto.Schema

  schema "users" do
    field :id, UXID, primary_key: true, autogenerate: true, prefix: "usr"
  end
end


# Decoding UXIDs (new)
{:ok, decoded} = UXID.decode("usr_01h2vx8qf9bhqzd0kf6x8h4n8z")

[johnnyt]

@jacobparry I just updated a few things and you'll need to rebase (sorry about that).

The functionality of the Ecto.UXID module that was removed also lives in the main UXID module. My hope with that was that just using UXID for ecto related things would work across the board. You mentioned that you're using typed_ecto_schema and that is uses the Elixir type of the module, which isn't working.

I updated the UXID.t() to now be a String.t(), and introducted a UXID.Codec module that now holds the expanded struct and type.

Could you check to see if using this GitHub repo in your mix.exs instead of the hex package works with 'typed_ecto_schema`?

[jacobparry]

Your idea of just changing the .t() function to map to a string is good. I had thought of that but didn't know if you had typed the struct to it specifically for a reason.

I updated my PR and took out the extra ecto_type stuff since that is redundant now with the change of the type string.

I left all the decoder work in and extra tests and everything.

[johnnyt]

I just updated a couple of things (including formatting), could you rebase again? I'd love to get this merged in.

[johnnyt]

Looks good - thanks for this!