v0.13.0

Thanks to @joaohbbittar for the detailed report in #26, MyoGen 0.13.0 corrects the Fuglevand force model and the Hill muscle activation.

• Force model: the rate-dependent twitch gain is computed per discharge from the preceding inter-spike interval, and twitches are summed exactly. Force no longer depends on recording_frequency__Hz, saturates with discharge rate, and depends on contraction time.
• Twitch-force range: the new optional twitch_force_range sets RP independently of the recruitment range, as in Fuglevand et al. (1993).
• Hill muscle activation: summed activation now saturates at 1 instead of about 2.5 with the FDI defaults, so isometric muscle force stays below F0.
• Current injection: inject_currents_and_simulate_spike_trains now works in a fresh interpreter.

Upgrade notes: absolute ForceModel outputs change. Force is now in units of peak twitch force P. Regenerate stored force baselines and re-tune %MVC targets. HillModel and SimulationRunner force and torque scale down, by about 2.4× at full saturation with the FDI defaults. Re-calibrate anything tuned against the old output.

ForceModelVectorized remains available as a subclass of ForceModel. The old inter-pulse-interval helpers in force_utils have been removed.

Full changelog