v1.150.0

Added

• bn.sprt decides whether a test is flaky with Wald's sequential probability ratio
  test. Give it a pass/fail callable and a bn.SprtCfg(p0, p1, alpha, beta, max_runs);
  it runs until the evidence says PASS or FLAKY (or INCONCLUSIVE at the cap) and
  keeps the error rates within Wald's bounds alpha/(1-beta) and beta/(1-alpha),
  unlike re-checking a Clopper-Pearson interval after every run. bn.Sprt takes one run
  at a time for callers that own the loop, and Sprt.to_svg() draws the walk as a
  standard-library-only scatter chart against the PASS and FLAKY lines. The result
  lists the run index of each failure, and SprtCfg.runs_to_pass, flaky_window and
  expected_runs give the run counts to budget for. bencher/example/example_sprt.py
  sweeps the true failure rate and plots the run count and the FLAKY rate (#1216).