The member types `phase_label` and `phase_sweep_policy_type` formerly
required a definition as part of the TKSVM simulation class interface.
The need for this historically originated in when classifiers and sweep
policies were determined at compile time. Now that these are derived of
abstract policies, the remaining uses of these types could be replaced
by their canonical definitions:

    using phase_label = typename phase_space::classifier::policy<phase_point>::label_type;
    using phase_sweep_policy_type = phase_space::sweep::policy<phase_point>;

TKSVM-compliant simulation classes are thus now longer required to
implement these.

This enables simulation classes to intentionally return an "empty"
configuration in cases where no samples are desired but the simulation
is merely run as part of a larger PT set.

"Emptiness" is established by calling empty() on the configuration.
Since not all configuration type may implement empty(), the comparison
`size() == 0` is a compile-time fallback. If no `size()` is provided,
the distance between begin and end iterators is used.

Derived config policies in client codes may want to override it.

Introduced two new ClusterPolicies which adapt `lattice::bravais` for
cluster iteration:
* `frustmag_lattice` (selected by parameter cluster=lattice) uses the
  natural bravais lattice unitcell as the spin cluster
* `frustmag_single` (cluster=single) uses a single-spin cluster

This fixes a problem when invoking frustmag-learn without MPI. Since
there is only one process (rank == 0), it tries to read the permutation
for all clones.

Renamed concept of `LatticePolicy` to `ClusterPolicy` to avoid confusion
with frustmag `Lattice`s.

Define `gauge_config_policy` as an alias to `clustered_config_policy`.

Its configuration(...) function now adheres to the common method of
accessing elements by `cell[block(a)][component(a)]`.

See Appendix D of 1907.12322

Omitting the crossover regime (previously regime III in "Kagome2")

If the restored slice_measurements contained accumulators which are not
defined in the prototype, the previous (more "elegant") implementation
meight have thrown (depending on lexicographic ordering of accumulator
names). The new impl does allow for additional accumulators in the
restored slice_measurements. This situation may arise of subclasses
define additional variables (as is the case for the test_adapter).

Measures the contribution of the additional term of the learn rank-3
o.p. tensor as opposed to Zhirtomirsky's result. Does it encode the
constraint? We will see...

No idea why the case of single MPI rank communicators were excluded here
in the first place. The permutation is trivial ({0}), but the variable
still has to be initialized and cannot be empty.

Users can choose to use the lattice unit cell ("lattice"; previous
behavior; default) or single spins ("single") as the "cluster"
parameter. This mirrors the similar feature supported by the gauge model
config policy for a long time.

The `polygon` utility class can now handle "1d polygons" (a.k.a.
intervals) which can be used to specify 1d phase diagrams by assigning
phases to several intervals.