v0.2.0

• Added the 0.2 run knobs, as fluent setters and as environment variables:
  shrink()/shrinkBudgetMs() (report a counterexample as generated, or bound
  the descent), phases() (PROPERTY_PHASES), derandomize()
  (PROPERTY_DERANDOMIZE) and path() (PROPERTY_PATH, replaying a recorded
  shrink descent instead of searching for it again). Precedence follows one
  rule, now stated in the README: the environment dials the suite and wins for
  PROPERTY_RUNS/PROPERTY_PHASES/PROPERTY_DERANDOMIZE, while the code pins
  the property and wins for seed()/path().

• A property whose id was derived from a closure now says so on stderr, through
  the channel that already carries the excessive-discard warning. Such an id
  keys the regression corpus by something that moves — {closure} on PHP 8.3
  collapses every closure of a class onto one key, and {closure:file:line}
  from 8.4 moves when a line is inserted above. id() is the fix.

• Requires rasuvaeff/property-testing-core ^0.2. The knobs above are
  0.2 engine fields; there is no version of this adapter that offers them
  against core 0.1.

• Added PropertyCheck::id(): names the property, replacing the id derived
  from the calling method. The string is used verbatim — as the
  regression-corpus key, as the id on every event, and as the property's
  display name. It exists for the case the derived id cannot serve: a
  forAll() called from a closure, where PHP has no stable name to derive
  from. On PHP 8.3 every closure of a class is {closure}, so two properties
  in one file share a corpus key and overwrite each other's counterexample;
  from 8.4 the name carries a line number, so an edit above the property
  orphans yesterday's entry. Neither throws — the corpus simply stops
  replaying. Pest's it()/test() bodies are the common source of both.