Trajectory request signature

[o-murphy]

We plan to add a new get_trajectory method to the interface, which will allow more flexibility in requesting a trajectory compared to fire

We propose the following changes:

• filter_flags instead of extra_data (extra_data is no longer useful and should be deprecated)
• termination_flags - specifies which hardcoded termination events should be enabled, for example (stop on Zero)
• record_by - which axis or state attribute should be used for recording (range.x, time, velocity, look_distance, height, etc.)
• record_request - sortable iterable object for matching with record_by
• get_trajectory will handle or suppress Range Errors by default, guaranteeing return of a HitResult with complete/incomplete trajectory

[dbookstaber]

Termination conditions we want include every TrajFlag:

• Stop at zero (i.e., when crossing down through sight line)
• Stop at apex (i.e., when vertical velocity goes from positive to zero/negative)
• Stop at MRT (i.e., when slant-height begins to decrease)
• Stop at distance – x-distance, y-distance, slant-distance, slant-height

Also, it would be nice to temporarily modify something in the config for a single request. Like get_trajectory(..., minVelocity=Velocity.MPS(50)).

[o-murphy]

Termination conditions we want include every TrajFlag:

• Stop at zero (i.e., when crossing down through sight line)

• Stop at apex (i.e., when vertical velocity goes from positive to zero/negative)

• Stop at MRT (i.e., when slant-height begins to decrease)

• Stop at distance – x-distance, y-distance, slant-distance, slant-height

Also, it would be nice to temporarily modify something in the config for a single request. Like get_trajectory(..., minVelocity=Velocity.MPS(50)).

modifying config on the fly looks like wrong architecture pattern, maybe we should specify **kwargs for Mutable parameters instead of config or put modified config.copy() directly to the method.

[dbookstaber]

Actually we may want to repurpose the extra_data argument to work like SciPy's dense_output bool: In order to interpolate reliably for things like HitResult.danger_space we need the HitResult to store all BaseTrajData points.

[o-murphy]

This seems fair.