refactor transport

[ecoon]

The Transport_ATS is in need of a major refactor. See all issues tagged with reactive-transport-roundup.

A first cut at this is in review at #232 . This will serve as a running list of things still to be fixed from that PR.

• Several PK_DomainFunctions use fields, but do not require their data or evaluators. Currently some, but probably not all, of these fields and their evaluators are Required in transport_ats_pk. These should get moved to the domain function. Furthermore, most PK_DomainFunctions probably don't call Update() on the fields that they do use, which could easily introduce bugs. At least PK_DomainFunctionField does call Update if there is an evaluator -- see that implementation as an example -- though it does not call require on that field.
• Remove all references to inner subcycling e.g. search for: if subcycling_ and tag_subcycle_{next,current}_
• Figure out the difference between primary and secondary, aqueous vs solid vs non-advected vs gas, etc. For instance, does component_names include secondary species or just primary? What vectors should include solid_residue?
• Remove the many many local pointers to memory. Get the data from State -- these pointers may not stay constant in the future!
• Add interpolation evaluator, removing manual calls of saturation, porosity, density at NEXT and CURRENT

[levuvietphong]

The following regression tests set the (transport) inner subcycling flag true:

- 06_transport/surface_tracer.xml
- 06_transport/subsurface_tracer.xml
- 07_reactive_transport/dilution_test.xml
- 07_reactive_transport/subsurface_sorption.xml
- 07_reactive_transport/subsurface_sorption_logical.xm
- 07_reactive_transport/surface_decay_ingrowth.xml
- 07_reactive_transport/surface_denitrification.xml
- 07_reactive_transport/surface_denitrification_logical.xml

Should we update these tests and set the flag to false?