"B" is flagged to expand to "byte" or "bytes", but it very often actually means "billion"

**Describe the bug**

> because at 70B parameters that's way too large for most home computers!

The "B" to "byte(s)" linter doesn't seem like the right place to add the "B" to "billion" expansion.

We could have a dedicated one for "B" that offers both "byte(s)" and "billion" and does its best to decide which would be correct in context. But that might lead to inconsistent `message` wording between the "B" linter and the one that lints all other memory size abbreviations.

I'm not sure how reliable it is to have two linters flag the same thing. I've seen this but I know we have code to deal with "overlapping" lints which might prevent that now. We might need a way for linters (or lint groups?) to declare that they can overlap with each other?

The issue provides no file or test entry point. Start by locating the linter for byte abbreviations and the handling of overlapping lints; establish expected behavior for “B” meaning billion versus bytes and decide how competing diagnostics should be resolved.

Written by the indexing model from the issue text.