sedimentation removal dynamic (for 0-dimensional models)

[slayoo]

For the parcel model, removal would be based on:

prob = terminal_velocity * dt / length_scale = K * dt 

(which is a linearisation of a Poissonian formula)

where length_scale could be cbqr(parcel_volume)?
and K denotes the removal rate (see eq. A18 in Curtis et al. 2016)

then do Monte-Carlo removal comparing the prob. with u01

removal with the same flag as in SDM invalidation

test could check if this is robust to changing dt and n_sd

other option: multiply the multiplicity by the prob?

[slayoo]

CC: @tluettm

[slayoo]

let's track it here: #1722