Snowfakery

Fix regression when an SObject declares a RecordType on some records
but not others. 301

Add a new feature for integrating with CumulusCI's Load architecture.
This feature will obsolete most (hopefully all) uses of mapping.yml with Snowfakery. Documentation for it will be in CumulusCI. 290

Adds support for Salesforce Person Accounts to Snowfakery.
See the documentation for details. (283)

Fix regression: Can set RecordType on objects with names that
are SQL keywords (like Case). (277)

Record types can now be specified with `random_choice`. 273

Plugins are now searched for in relative directories called "plugins"
as well as a directory under the user's home directory. See the docs
for more information. 229

Snowfakery can now generate SQL text as an output format. The specific
schema is compatible with CCI's data loader, but can be used in other
contexts. 162

Add a `snowfakery_filename` variable to allow injecting the
recipe filename into recipe output. 249

Add NULL to Snowfakery formulas 239

Allow references to be indirect and include dots in them.
Consequently, dots and other punctuation are now disallowed in
nicknames and discouraged in object/table names. 186

Snowfakery's default branch is now main, so references to
master should be updated.

Adds a "random_reference" feature to allow randomized
connections between objects as described in the documentation.
(PR 184)

Adds a notion of "variables" as described in the documentation. These allow computed values to be easily shared between object templates.
(PR 231)

Adds a "snowfakery_locale" variable which determines what locale drives fake data creation (person nammes, state/province name etc.)
(PR 231)

Does not depend upon pygraphviz library which can be difficult to install. One will still generally want to install
[graphviz](https://graphviz.org/), but not both graphviz AND pygraphviz. (PR #171)

Calls to "fake" are optimized. Recipes that are heavily reliant on faked fields may be 10-15% faster. One tradeoff is that e.g. person names or state names do not follow a real-world statistical distribution. Rare names (in en_US locale) like "Mathias" are now as common as real-world common names like "Matthew". (PR 214)

Adds Salesforce Example Files.

Adds an official API for embedding in Python.

Updates dependencies including Faker, which has Salesforce-contributed optimizations.

Adds min, max, round to the math plugin.