revised exposure factor implementation for bright time

[changhoonhahn]

py/desisurvey/ephem.py

• sun_ra and sun_dec added to ephem table

py/desisurvey/etc.py

• bright_exposure_factor function calculates the bright time exposure factor from airmass, moon illumination, moon separation, moon altitude, sun separation, and sun altitude as input.
• _bright_exposure_factor_notwi function calculates the bright time exposure factor without twilight contributions from airmass and moon parameters. The function uses a polynomial regression model fit to exposure factors calculated using an improved sky brightness model (see notebook). The model is based on re-fitting Krisciunas & Schaefer (1991) to BOSS sky fiber data plus twilight contribution from Parker Fagrelius's thesis work. See notebook for a comparison of the original KS1991 sky model to refit+twilight model.
• _bright_exposure_factor_twi function calculates the twilight contribution to the bright time exposure factor. The function uses a simple linear regression model that takes airmass, sun separation, and sun altitude as input.

py/desisurvey/scheduler.py

• next_tile method: If use_brightsky==True, bright time exposure factor is included in next tile selection
• update_exposure_factor method implemented. This method is also called in surveysim.nightops.simulate_night to correct the ETC on the fly.

[dkirkby]

Let me know when this is ready for review. Note that @schlafly is working on #108 in parallel.

[schlafly]

I think these will be pretty orthogonal, fortunately!

[changhoonhahn]

@dkirkby I think this pull request is ready for review. I've checked that it runs expectedly in conjunction with the pull request I submitted for surveysim. This pull request should also address the issues that @sbailey pointed out in the previous pull request I submitted.