ex-fuzzy 3.5.0

Added

• Ordinal conditions. ordinal_conditions=True (or a collection of features) lets a rule say x IS at least Fair or x IS at most Fair over a variable's ordered terms, besides a single term. A variable with n terms offers 3n − 4 conditions. The membership is the bounded sum of the terms covered, which on a strong partition is a shouldered trapezoid.
• Named compound terms. compound_terms={feature: [(name, [terms]), ...]} adds conditions such as local_hour IS Night, a bounded-sum union of the listed terms that may wrap around (for example evening ∪ night across midnight).
• Both work with BaseFuzzyRulesClassifier (winning-rule and mixture inference, also with mixture_search) and BaseFuzzyRulesRegressor. They print readably, and rule.conditions() exposes each condition's operator (is, at_least, at_most, any_of). Text rules, variables and pickles round-trip.

Both options are off by default, and then every fit is bit-identical to 3.4.0 (nine recorded fits are replayed in the tests). See docs/source/ordinal_conditions.rst and benchmarks/benchmark_ordinal_conditions.py.

Full changelog: v3.4.0...v3.5.0