[Quantization] Re-consolidate MXFP8 and FP8 linear converters

[aditvenk]

More info in #3199 (comment)

[aditvenk]

@acisseJZhong -- assigning to you based on comment thread

[tianyu-l]

The title seems inaccurate -- the consolidation should happen for mxfp8 linear and mxfp8 grouped mm

[tianyu-l]

should migrate to pytorch/ao#4374