Update non-peak function fitting test instructions

[RichardWaiteSTFC]

Describe the outcome that is desired. It has been found during smoke testing (#38241 and #38217) that fitting an exponential decay to the EMU data in the instructions is quite flaky/difficult https://docs.mantidproject.org/nightly/tutorials/mantid_basic_course/fitting_data/03_fit_model_choices.html

The starting parameters are quite bad (in particular the ExpDecay Height which should be ~40000) and the default LM minimser fails to converge on a good fit (seen on windows and IDAaaS).

Describe any solutions you are considering It was decided that it would be useful to advise the user to change the initial starting parameters, and perhaps also minimiser (Simplex seems to work well, even from bad starting parameters).