Model NIRCam MASKLWB offsets to cross-check handling of broadband filters

[mperrin]

See discussion on #509

@JAStansberry :

The use of a different (longer) effective wavelength for calculating the offsets for the W filters came from John Krist, based on modeling he did.
It would actually be useful to use the current version of WebbPSF to model the contrast as a function of displacement along the bar. John was using a pretty notional OPD map, I think.

Also, looking back at my script for calculating the offsets, the were based on pixel scales of 32 and 65 mas. My guess is that the difference due to that probably can't be measured, but we may as well use the right values. Even better, use the scale appropriate to the location of the bar masks and taking into account the new distortion model for the coronagraph FOV.

@mperrin:

@JAStansberry and @JarronL, we could indeed do some contrast calculations for different MASKLWB offsets as you propose. (I'm about to finally update the OPD maps for as-built latest & greatest, too, so these would be much better than earlier notional OPDs). That's probably better done as a task by Jarron or someone else on the NIRCam IDT rather than any of us in the webbpsf or telescope teams - I'd be happy to advise on how to do the modeling though. It sounds like the choices John Krist made may have been motivated by getting as deep as possible contrast, at a cost of inner working angle. Which may well be the right call but it's something to clarify and convey to users.

@JarronL:

Sounds good. @mperrin, if you could make a new PR and tag me as the responsible, I can look into it ::waves hand vaguely:: sometime in the near future. Would like to understand how the most recent filter-dependent bar mask reference positions as defined in SIAF were calculated, though, which probably requires some combination of @mgennaro, @JAStansberry, @juliengirard.