`sig_tf_r_max(1)` non-idempotent in ST problems

[jonmaddock]

sig_tf_r_max(1) appears to be non-idempotent when trying to evaluate the solution vector in ST cases. It should, of course, be stable after a maximum of 10 model evaluations. The warnings in #3351 will help identify when this is happening.

[ajpearcey]

Using an IN.DAT for ST device without using f-values I cannot reproduce these non-idempotent solutions. It no longer produces any warnings for non-idempotent solutions.

Therefore I will close this issue.