Feature Request: 8 bit fixed point numbers

I'm interested in using 8 bit fixed point numbers in jax and numpy in the range [0, 1], and [-1, 1]. I think with the current naming convention it would be something like `float8_e0m8` and `float8_e0m7`.

The issue names no files or tests. Start by reviewing the existing dtype extensions and naming convention, then clarify the representation and expected behavior for [0, 1] and [-1, 1]; done requires an agreed design and coverage for both 8-bit fixed-point types.

Written by the indexing model from the issue text.