Ambiguities introduced by parametric subtypes

[alecarraro]

The addition of parametric subtypes in PR #332 introduces some ambiguities and seems to require further refactoring. I’ve noticed at least two problems so far:

1. Ambiguous categorization of systems with parametric matrices

A system of the form

$$x' = A x$$

where A is an interval matrix can naturally be categorized as a parametric system.
However, in MathematicalSystems it is also possible to construct a LinearContinuousSystem with an IntervalMatrix, since IntervalMatrix is a subtype of AbstractMatrix.

This creates an ambiguity

2. _corresponding_type does not distinguish parametric vs. non-parametric

The internal function

MathematicalSystems.jl/src/macros.jl

Lines 151 to 161 in 0655e43

	function _corresponding_type(AT::Type{<:AbstractSystem}, fields::Tuple)
	TYPES = types_table(AT)
	FIELDS = fields_table(AT)
	idx = findall(x -> issetequal(fields, x), FIELDS)
	if isempty(idx)
	throw(ArgumentError("the entry $(fields) does not match a " *
	"`MathematicalSystems.jl` structure"))
	end
	@assert length(idx) == 1 "found more than one compatible system type"
	return TYPES[idx][1]
	end

only considers field names for classification, not the types of the fields.
As a result, a LinearContinuousSystem and a LinearParametricContinuousSystem (which share the same field names) are indistinguishable here.