Fix AudioParam behaviour

[b-ma]

Hey, some new stuff regarding AudioParam

• Fix AudioParam argument parsing
• Fix AutomationRate set behaviour on control thread
• ~Fix return value for raw setters (i.e. set_value and set_automation_rate)~~
• Fix return value for raw set_value setter
• Some refactoring

[orottier]

Thanks for the changes.
I will probably pick this PR apart in the separate issues

• analyser panics, to be continued in fix: unwanted panic at Analiser initialization #422
• validation of audio param numerical values
• the issue of automation_rate_synchronicity (needs work)

[orottier]

I don't think marking these methods as &mut self will work in practise, because they are return as immutable refs from the AudioNodes:
pub fn gain(&self) -> &AudioParam

This could of course be changed to pub fn gain(&mut self) -> &mut AudioParam but that is a rather big change where we should assess pros and cons.

[orottier]

Including the non-Arc field here breaks the Clone implementation (which we need for the node bindings).
Calling set_automation_rate on a cloned AudioParam will not update the automation rate of the original AudioParam

[b-ma]

Hum right, I didn't think about these points indeed (and sorry I indeed forked from the previous PR branch rather than from main)

For the automation rate problem, maybe the more straightforward solution is then just to update it through AudioParamShared when calling the method on control side rather than when the message is handled on render side, as in any case the renderer only uses it own automation_rate copy?

I let you cherry pick the changes, feel free to close the PR when needed

[orottier]

Continued in #425 #422 and new issue #426