Muon asymmetry fitting suggestions

Suggestions from JL following testing of https://github.com/ISISComputingGroup/ibex_bluesky_core/issues/227:

- On failed fits, return `NaN` as fit parameters (not actually sure this is a good idea...)
* Requires some work to 'chase through' various consumers who will be upset by NaNs, e.g. `lmfit` at outer level will need `nan_policy="omit"`
- Want to expose a "figure of merit" which is `(asymmetry)^2 * (count rate)` from `MuonAsymmetryReducer`
* Should be relatively easy to expose this from `MuonAsymmetryReducer`
* If the asymmetry fit fails on a low stats run, the asymmetry shouldn't exceed 1 anyway so a good substitute is just to put Figure of Merit=rate in this case. Assuming other scan points have a large flux and fit successfully, it'll be close to the baseline for the outer peak fit.
* For the two-oscillation function there will be figure_of_merit_0 and figure_of_merit_1 derived from A_0 and A_1 respectively. I'll want to choose to plot/optimise one of these.

- Expose `r_squared` from `MuonAsymmetryReducer`.
- Give more decimal places on the soft signals that hold fitting results
- Once https://github.com/isiscomputinggroup/ibex/issues/8809 is done, read $t_0$ and $t_{good}$ PVs and use these to set the start of the data rebinning range (this is an adjustment in the inst scripts side, not in `ibex_bluesky_core`).
* Adjust all DAE time coordinates by subtracting $t_0$, then set rebinning to start at $t_{good}$