Implement variable number of accretion species and ensure proper treatment of accreted material in binaries

[This issue was originally posted to Assembla by Pablo.]

In star/public/star_def.inc, we had a hard-coded number of accretion species set by `max_num_accretion_species = 50`. As a workaround this was increased to 250 by Josiah and it will probably take a while before someone hits that limit while modelling an interacting binary. This causes issues in binary systems with large nets. Ideally, the pointers specifying the composition of accreted material should have variable sizes, and the binary module should take care of updating for both stars such that its size matches that of the star with the larger net.

In addition, the treatment of accreted material when the stars have different networks is not transparent. Perhaps options need to be put in place in binary to ensure both stars include the same isotopes in their networks.

Start with star/public/star_def.inc and trace the binary module's handling of accreted material for both stars. Determine how variable-sized composition pointers and differing isotope networks should be coordinated; the work is done when the species size follows the larger stellar network and accreted material is treated consistently.

Written by the indexing model from the issue text.