Deprecate Phase::name in favor of Phase::id

Currently, there is a duplication of Phase::name() and Phase::id(), where the former is supposed to be unique, whereas the latter is consistent with the phase id specified in the input file. In current Cantera, however, the context is always clear, so the differentiation is not needed.

A differentiation of name and phase_id does make sense for Solution objects in the Python interface (and potentially serialization of Solution in C++). To preserve this capability, the name is reassigned to the C++ object Solution (see #696).

Further:

Currently, phase name/id can be used when accessing species within a Phase, i.e.

 * A species name may be referred to via three methods:
 *
 *    -   "speciesName"
 *    -   "PhaseId:speciesName"
 *    -   "phaseName:speciesName"

The latter are, however, not used. I.e. per @speth's comment in #696

I believe that these latter two methods are not actually in use within Cantera, and I'm really not sure that we need or want them. For multiphase mixtures (C++ class MultiPhase, Python class Mixture, you always need to provide both a phase name/index as well as species name in order to look up information about a species in a particular phase. And for surfaces, the usual approach is that the species names are unique among the phases, e.g. H2O is a gas species and H2O(s) is the same species absorbed onto the surface. So at some point, I was planning to deprecate and remove this capability anyway.

Changes proposed in this pull request:

• Replace references to Phase::name by Phase::id in equilibrium calculations and thermo phases
• Deprecate alternate access methods for SpeciesIndex in Cantera 2.5

Pending

At this point, the only instance where Phase::name is still required is the Python interface (several tests check name rather than ID). Once #696 is merged, name will be part of the C++ Solution object, i.e. Phase::name() can be deprecated.