This is part of a push to support larger files. The focus is the switch
to using Uint8Array to store binary data. But also includes:

- When running on Node, use Buffer.from() for hex string conversions.

- To avoid large buffer copy, signedMsg as been replaced by an object
    containing  iv, encrypted, and hmac.

- hmac calculation has changed so it avoids copying (possibly very
    large) encrypted data. See signDigest() in lib/codec.js.

- Minor cleanup

Handling hex encode/decode at the input/output boundaries and using
Uint8Array internally for representing binary data has these benefits:

- More memory efficient, allows processing of 2x larger files.

- Aligns with cryptographic best practices: hashing is now performed
  on raw binary data (Uint8Array) instead of hex strings.

- Behavior is (mostly) unchanged
  - scripts/index_template.html textContent is not implemented and
    needs to be redesigned.

makes unnamed function more self documenting

Also removed use of recursion to improve readability (and debugability).
As a bonus, function is actually shorter (LoC AND lines of text)

cuts size by 1/3 and noticeably improves performance

Refactored encrypted buffer handling to reduce memory usage
Minor cleanup