Update guess_tempo_parameters and tutorial on parameter choice / dkmax anomaly

[piperfw]

Addresses #96 and #23. Changes:

• guess_tempo_parameters can now use an optional system argument to determine a suitable timestep to resolve system dynamics; the smaller of this estimate and that made used analysing the bath correlations is used (see Improve guess_tempo_parameters depending on system parameters #96 for further discussion)
• improved coverage of tempo.py modules
• tutorial parameters.ipynb including
  • example convergence of tcut, dt, epsrel
  • physical (e.g. bath correlation decay timescale) and practical (e.g. epsrel heuristic) information to determine convergence, and a suggested approach (determine tcut at rough dt, epsrel, then check for convergence under dt,epsrel->0)
  • use of guess_tempo_parameters with and without a system argument
  • [style: axes of oqupy.plot_correlations_with_parameters are now centred]
  • less common arguments to TempoParameters (subdiv_limit, liouvillain_epsrel, add_correlation_time) and degeneracy checking (unique=True) in TEMPO computations
  • Considerations for PT-TEMPO, including the 'dkmax anomaly' (Tutorial to TEMPO parameter choice & the dkmax anomaly #23), which I was able to reproduce with a calculation that takes barely a minute at large dkmax, but quickly becomes infeasible when increasing dkmax

I would particularly appreciate feedback on the tutorial text @gefux, as this was written based on my own experiences of using TEMPO and what I have learnt from working with the parameters code over the last few days only.

[gefux]

@piperfw: Thank you very much for this.

The tutorial is phenomonal. This is can be really helpful in makeing OQuPy more accessible for users outside of the St Andrews group. Together with the much improoved parameter guessing function I think it should therefor be added to the list of major contributions in v0.5, if you agree.

I will hopefully find the time tomorrow to have a closer look at the parameter guessing funciton. I will get back to you on this then.

[piperfw]

High praise indeed! I don't see any harm in making the list of major contributions to v0.5 look longer :), thanks for the suggesting that.