v0.1.0

• Mining thresholds by freq or support: mine_corpus(min_doc_support=...),
  mine_document/mine_within(min_unit_support=...); all given thresholds
  combine, defaults unchanged when none given.
• Support (relative frequency) alongside the absolute freqs:
  Pattern.doc_support / unit_support, filter(min_doc_support=, min_unit_support=), and matching patterns_frame columns.
• Per-occurrence free/constrained classification exposed:
  Pattern.free_occurrences, Pattern.constrained_occurrences, a free
  column in occurrences_frame and n_free_occurrences in
  patterns_frame.
• Semantic change: patterns that never occur free — every occurrence
  enclosed in a free occurrence of a larger pattern (e.g. are you always
  inside how are you) — are discarded from mined inventories (dialign's
  free/constrained rule). Kept patterns still report all their occurrences.
• Corpus model (Corpus, Document) with text helpers (patternminer.nlp).
• Cross-document mining (mine_corpus) and within-document repetition
  mining (mine_document, mine_within) on pygstlib generalized suffix
  trees.
• Pattern / PatternSet result types with filtering and maximal().
• Containment hierarchy (PatternSet.hierarchy()): regroup subpatterns
  under parent patterns.
• Optional pandas layer (patternminer[analysis]).
• Tutorial notebooks (quickstart, David Copperfield corpus study) and full
  README, both output-asserted by tests.