Fix completion stopping after a boolean flag (#142)

[jcruizcue]

A boolean flag (is_flag=True) consumes no value, so matching it as the last token must not short-circuit completion of the remaining parameters. Previously the completer returned nothing after a flag; now it continues offering the command's other options and arguments while still not offering true/false for the flag itself.

• _completer.py: skip the param_called/break short-circuit for flags
• tests: assert completion continues after a flag (options + arguments)
• Changelog: add entry

Fix #142

[auvipy]

@charan-rathore would you mind trying and reviewing it?

[charan-rathore]

Tried it locally. On master, completing after a boolean flag returns nothing: cmd --verbose <TAB> offers neither the remaining argument choices nor the other options. On this branch both complete, and no true/false value is offered for the flag itself. The new tests cover the short and long flag spellings, the --shout/--no-shout form, and more than one flag in a row; the full suite passes here (72 passed, 5 skipped). Looks good to me.

[auvipy]

Looks good to me.

then why the tests are failing in the CI?

[charan-rathore]

Checked the failing job: it is the flake8 tox env, not the tests (click81/82/83 all pass). One lint error: tests/test_completion/test_common_tests/test_option_completion.py:59 is 93 chars, over the 90 limit (E501). I reproduce it locally with flake8 click_repl tests. Wrapping that one line should turn CI green; the remaining cancelled jobs were cancelled by the matrix after the failure, not run to a real result.