Support for additional IRIS tracking modes

[tnakamot]

This pull request includes the following changes:

• Added new event coldStopMaskAngle. This was previously pupilRotation, but it was renamed because the cold stop mask angle can be different from the pupil rotation in special tracking modes (e.g., M1 tracking and DM0 tracking modes).
• The old pupilRotation remains because the IRIS DRS subscribes to it. Tracking attributes (pointingstate and trackID) are now removed because pupilRotation won't be used for tracking anymore.
• Revised the description of imgAtmDispersion and instrumentRotatorAngle to clarify that the TCS sends the correct angle taking into account the current tracking mode (normal tracking mode, M1 tracking mode or DM0 tracking mode).
• Revised the description of imgAtmDispersion and coldStopMaskAngle because the IRIS software needs those demands 3 seconds before the time indicated by the timestamp attribute.

[edwardchapin]

Other than requesting that @lianqiw confirm that we still need pupilRotation for the RPG, this PR looks good to me. We need the new coldStopMaskAngle event, and the updated Discussion clarifies things.

[lianqiw]

The RPG does need pupilRotation to properly rotate gradients from IRIS to NFIRAOS focal plane. This is not related to the Lyot mask.

[tnakamot]

After some discussion with Ed, I made some modifications to this pull request. Now this pull request includes the following changes:

• It is now clarified that "pupilRotation" may not be the same as the parallactic angle in special tracking modes (e.g., M1 tracking and DM0 tracking modes).
• Revised the description of imgAtmDispersion and instrumentRotatorAngle to clarify that the TCS sends the correct angle taking into account the current tracking mode (normal tracking mode, M1 tracking mode or DM0 tracking mode).
• Revised the description of imgAtmDispersion and coldStopMaskAngle because the IRIS software needs those demands "1 second" before the time indicated by the timestamp attribute. Added "TBC" mark to "1 second" because a concern about the accuracy of the future demand was raised.
• Increased the rate of imgAtmDispersion from 1 Hz to 2 Hz.

[edwardchapin]

This looks better now having reverted to using pupilRotation, but with the extra description to clarify things.