add in altitude support and simplify adding detector function

[ahnitz]

• Adds in the ability to set the altitude angles for custom detectors
• Simplify detector class, use lal only for its detector location information rather than the response value itself. This makes it easier to provide consistent metadata between custom and lal-builtin detectors
• Simplify the add detector on the earth function (use matrix multiplication operator, make the rotations clearer / more obvious / start rotations from simpler form of the basic arm response)

Todo

• update unittest to include both use of custom detector and crosscheck response matrix calculation against builtin lal value

[titodalcanton]

I am unlikely to be able to review this carefully before the end of the month…

[spxiwh]

I'm fine with adding this altitude angle in. I do want to check what this changes now. For e.g. if someone were doing some ongoing analysis, will detector responses change by this? (I think that we will expect CE to have changes, but not others?)

[spxiwh]

Why is this being changed?

[ahnitz]

@spxiwh The lal code doesn't actually use these numbers for much, but it does have them set properly to the right values. This change means that they will be set in the same way as the internal cached detectors in lal in case lal every does use them (for example if someone were to add high-frequency corrections to lalsuite it might matter). Before, I was not bothering to really set them because in the code itself the only thing it used them for was was a guess to a filter length.

[spxiwh]

So there's changes in here. I think to include the "altitude angles" of the two arms. Have you checked/can you show that this is unchanged in the limit that the altitude angles are both 0.

For current instruments (H1,L1,V1,K1), will this make any actual change?

[ahnitz]

@spxiwh That's actually what I added as a unit test. The key thing is that this sets up the response matrix. I numerically compare the response matrix here against the cached ones in lal. There is no change to the response there (except the E* case which has a 10^-4 level change due to the issue I raised in slack).

The altitudes are near zero for the H1/ L1 / V1 comparison, but not exactly so (nor should they be). In fact if you don't account for the altitude the response matrix would also be different at the O(10^-4) level like E* is.

[spxiwh]

Okay, I'm happy with this then.