move nsatz to MathComp -Analysis?

[affeldt-aist]

There is an instance of nsatz with rcfType in
https://github.com/affeldt-aist/coq-robot/blob/master/ssr_ext.v

Similarly, there is an instance of nsatz with realType in
math-comp/analysis#383

Should the rcfType instance of nsatz move in Analysis too? Or in MathComp? @CohenCyril @thery

[thery]

in math-comp/analysis#383 we could use the new ring of Kazuhiko but I didn't want to add a new depency.
For coq-robot, we actually use the grobner basis computation but maybe with some work this could be also a simpl ring I need to investigate.