Name of estimate instructions *e7 -> *a7 ?

[kasanovic]

From David Horner on merged pull request #557 :

vfrece7 - Float RECiprocal Estimate 7bits
vs
vfrecip7 Float RECIProcal estimate 7bits

I suggest we drop the e because of the association with element width. The 7 implies estimate.

I was also thinking Multiplicative INVerse Estimate 7bits vfminve7 if the e for estimate is deemed critical, but minv looks too much like minimum ?

anyway minor point, but I would

[kasanovic]

From Andrew Waterman:
@kasanovic - @David-Horner makes the valid point that the "e7" at the end looks awfully like the "eN" suffix we use on loads/stores to denote element width. TG can decide whether to keep current name, drop the "e", or replace it with "a" for "approximation".

[kasanovic]

I agree vfrec7 a nd vfrsqrt7 are clearer names.