Fixes to the amplitude noise implementation

[HGSilveri]

• Adds information of the beam's propagation direction to Channel
• Makes finite-waist effects depend on the position relative to the optical axis
• Makes amplitude fluctuations occur from run to run instead of pulse to pulse
• Fixes the transfer of laser_waist between NoiseModel and SimConfig
• Starts adding the pulser_version to serialized devices

Note: To avoid disruptions for serialized devices between different versions, propagation_dir is defaulting to None in all channels and QutipEmulator is assuming using propagation_dir=(0,1,0) whenever it encounter None. In a future version, it would make more sense to change the default propagation_dir of Global channels to be (0,1,0) and remove this assumption from QutipEmulator.

Fixes #712

[a-corni]

LGTM

[a-corni]

Nit: I am not a fan of "leaving", what about

Suggested change

	propagation_dir: The propagation direction of the beam leaving the
	propagation_dir: The propagation direction of the beam (or wave) associated with the

[a-corni]

Ah I forgot about one question: should we raise a NotImplementedError if the propagation vector is implemented in a channel that is not Rydberg.Global ?

[HGSilveri]

Ah I forgot about one question: should we raise a NotImplementedError if the propagation vector is implemented in a channel that is not Rydberg.Global ?

I think that is overly restrictive, how about I do that only for Local channels?

[a-corni]

LGTM !