anomaly with incompatible rewriting in clones

[alleystoughton]

Running

theory T.

op x = 4.

end T.

clone T as T' with
  op x <- 3.

gives the error

anomaly: EcLib.EcTheoryReplay.CoreIncompatible

Because errors about rewriting in clones are all localized at the
keyword clone, as opposed to the offending rewriting, this anomaly
needs to be turned into an error message that says (by analogy with
the error for type rewriting):

operator `x` incompatible operator declaration

or better

incompatible rewriting for operator `x`

[strub]

Yes. Error messages for clones are really bad (if non existent). We are now in a state where we can write better error messages.