run-tests: always define a custom-bin directory We are always generating `python` / `python3` executable and we will soon often generate a `hg` executable. So let's always make it a thing. This will help us to make sure we have the appropriate binary available to all script. On both linux and Windows. Differential Revision: https://phab.mercurial-scm.org/D11044

run-tests: explicitly track that `pythondir` was inferred If we want to know that something happened, we should simply track that this things happened. This will be useful to start using `tmpbindir` more in a different form. Differential Revision: https://phab.mercurial-scm.org/D11043

run-tests: use a global WINDOWS constant instead of multiple tests This should make the code clearer. This required the adjustement of a hack in the code testing this code. Differential Revision: https://phab.mercurial-scm.org/D11041

test-run-test: use explicit variables to clarify paths Since $TESTTMP and $TESTDIR are defined for both the main scope and the test run during the test, the situation get confusing. To work around that, we define some new, explicit, variables with the values we needs to keep track of. Differential Revision: https://phab.mercurial-scm.org/D11034

run-tests: also catch double-escapes for $TESTTMP On Windows the `$TESTTMP` string "repr" is different than its value, because `\` become `\\`. We introduce a new `$STR_REPR_TESTTMP` substitution to match that too. Differential Revision: https://phab.mercurial-scm.org/D11001

ci-windows: introduce manual windows CI Note: most of the foundational work of this commit was done by Matt Harbison, but enough has changed that I don't feel comfortable sending this patch as his. In our quest to remove Python 2 from Mercurial, we need to get Windows CI going to check that Python 3 support is up to an acceptable standard. This is the first step: adding a manual step to run a full Windows CI, with a certain definition of "full" since some things are not installed yet, like SVN. Differential Revision: https://phab.mercurial-scm.org/D11025

dirstate-item: deprecate tuple access on the class This should help us to catch and update the last user of this, especially in extensions. People will need to run the test with --pure to actually catch it, but this is better than nothing. Differential Revision: https://phab.mercurial-scm.org/D10992

dirstate-item: use the v1_serialization method in debugstate I am assuming the debug command are looking into the serialized format so I am using the `v1_` variants. This assumption might be wrong. Differential Revision: https://phab.mercurial-scm.org/D10988

dirstate-item: also build DistateItem in dirstate.directories() The rust code was building tuple. Differential Revision: https://phab.mercurial-scm.org/D10983

parsers: move DirstateItem to attr.s This is the followup I requested on D10949 - it was easy, so I just did it. Differential Revision: https://phab.mercurial-scm.org/D11030