Feature light lut

[YifanC]

• Use lut t0 profile to sample instead of using the average
• Use measured SiPM model instead of the RLC model

• Use pre-caclulated time profile instead of calculate it on the fly (This part should be static). You can see the differences of light_sample_inc (the direct outcome of this change). Note there's no randomness up to this step. I'm not sure this is hits the precision limit? Compare to the current way how we use LUT, it's not faster... Testing on 3 events, this PR takes ~145s, while using the current LUT takes 125s. I can factor this out in the PR if necessary.

• Not add first photon arrival time if not threshold triggered
• Only allow one trigger per event