Cumulusci

Fixed a problem building the Homebrew formula for installing CumulusCI
3.12.0.

Changes:

- We've removed the prompt that users see when trying to use a
scratch org that has expired, and now automatically recreate the
scratch org.
- The `load_dataset` task now automatically avoids creating Bulk API
batches larger than the 10 million character limit.
- Robot Framework:
- When opening an org in the browser, the Salesforce library now
attempts to detect if the org was created using the Classic UI
and automatically switch to Lightning Experience.
- The Salesforce library now has preliminary support for Summer
'20 preview orgs.
- CumulusCI now directs `simple-salesforce` to return results as
normal Python dicts instead of OrderedDicts. This should have
minimal impact since normal dicts are ordered in the versions of
Python that CumulusCI supports, but we mention it for the sake of
completeness.

Issues closed:

- Fixed an issue where non-ASCII output caused an error when trying to
write to the CumulusCI log in Windows. (1619)

Changes:

- CumulusCI now includes
[Snowfakery](https://pypi.org/project/snowfakery/), a tool for
generating fake data. It can be used to generate and load data into
an org via the new `generate_and_load_from_yaml` task.
- Added two new preflight check tasks for use in MetaDeploy:
`get_available_licenses` and
`get_available_permission_set_licenses`. These tasks make available
lists of the License Definition Keys for the org's licenses or
PSLs.
- The `get_installed_packages` task now logs its result.
- Robot Framework: Added two new keywords (`Get Fake Data` and
`Set Faker Locale`) and a global robot variable (`${faker}`) which
can be used to generate fake data using the
[Faker](https://pypi.org/project/Faker/) library.

Issues closed:

- Fixed an error when loading a dependency whose `cumulusci.yml`
contains non-breaking spaces.
- Fixed a PermissionError when running multiple concurrent CumulusCI
commands in Windows. (1477)
- Show a more helpful error message if a keychain entry can't be
loaded due to a change in the encryption key.
- Fixed the `org_settings` task to use the API version of the org
rather than the API version of the package.
- In the Salesforce Robot Framework library, the `Open App Launcher`
keyword now tries to detect and recover from an occasional situation
where the app launcher fails to load.

Changes:

- Added `custom_settings_value_wait` task to wait for a custom setting
to have a particular value.
- The `metadeploy_publish` task now has a `labels_path` option which
specifies a folder to store translations. After publishing a plan,
labels_en.json will be updated with the untranslated labels from the
plan. Before publishing a plan, labels from other languages will be
published to MetaDeploy.

Issues closed:

- Fixed an issue where running subprocesses could hang if too much
output was buffered.

Issues closed:

- The `batch_apex_wait` task will now detect aborted and failed jobs
instead of waiting indefinitely.
- Fixed reporting of errors from Robot Framework when it exits with a
return code \> 250.
- Fixed an ImportError that could happen when importing the new
metadata ETL tasks.
- Fixed bugs in how the `set_organization_wide_defaults` and
`update_admin_profile` tasks operated in namespaced scratch orgs.
- Show a more helpful error message if CumulusCI can't find a
project's repository or release on GitHub. (1281)
- Fixed the message shown for skipped steps in `cci flow info`.
- Fixed a regression which accidentally removed support for the
`bulk_mode` option in bulk data mappings.

Critical changes:

- The `update_admin_profile` task can now add field-level permissions
for all packaged objects. This behavior is the default for projects
with `minimum_cumulusci_version` \>= 3.9.0 that are not using the
`package_xml` option. Other projects can opt into it using the
`include_packaged_objects` option.

The Python class used for this task has been renamed to
`ProfileGrantAllAccess` and refactored to use the Metadata ETL
framework. This is a breaking change for custom tasks that
subclassed `UpdateAdminProfile` or `UpdateProfile`.

- Refactored how CumulusCI uses the Bulk API to load, extract, and
delete data sets. These changes should have no functional impact,
but projects that subclass CumulusCI's bulk data tasks should
carefully review the changes.

Changes:

- New projects created using `cci project init` will now get set up
with scratch org settings to:
- Use the Enhanced Profile Editor
- Allow logging in as another user
- _Not_ force relogin after Login-As
- If `cumulusci.yml` contains non-breaking spaces in indentation, they
will be automatically converted to normal spaces.
- Bulk data tasks:
- Added improved validation that mapping files are in the expected
format.
- When using the `ignore_row_errors` option, warnings will be
suppressed after the 10th row with errors.

Issues closed:

- The `github_release` task now validates the `commit` option to make
sure it is in the right format.
- If there is an error from `sfdx` while using the `retrieve_changes`
task, it will now be logged.