payroll-type — US:403b:Employee:Roth — needed separate since taxable

Since Roth contributions are taxable, there are some reports that
need to include these amounts in total salary (i.e., when running a
report that seeks to show total taxable income for an employee). As
such, we need a `payroll-type` specifically for Roth 403(b)
contributions.

Add `tax-implication` metadata type of Asset-Sale

This will be used for sale of stocks, for example.

Add new payroll type codes

Oregon added a new payroll tax for disability so we need to add the
payroll-types for Ohio's state and local taxes.

reconciler: Fix reconciler lumping unmatched books items together (#20737)

Issue was that the subset matching was passing through the grouped transactions
rather than the original transactions when it failed to match.

reconcile: Update to support standard FR CSV format

It seems as though we were previously using a transaction search export CSV,
rather than the official statement CSV.

reconcile.helper: Avoid rt >= 3.0 library due to breaking changes

Error relates to rt.Rt not existing.

reconcile: Make bank-statement path relative to $CONSERVANCY_BEANCOUNT.

Also add CSV metadata.

Add typing dependencies for requests, dateutil, setuptools and yaml.

Without these, mypy complains that "Library stubs are not installed".

config: Fix mypy error for git.exc.

Error was:

conservancy_beancount/config.py:142: error: Trying to read deleted variable "exc"

Not entirely sure what's causing this, but since GitPython also imports the
exceptions at the root of the package with "from git.exc import *", we can
import them from there instead, which seems to appease mypy.

reports: Fix mypy errors.

Errors were:

conservancy_beancount/reports/core.py:923: error: Generator has incompatible
item type "Optional[str]"; expected "str"

conservancy_beancount/reports/core.py:929: error: Item "None" of
"Optional[str]" has no attribute "lower"

conservancy_beancount/reports/ledger.py:534: error: Item "None" of
"Optional[str]" has no attribute "partition"

conservancy_beancount/reports/ledger.py:729: error: Item "None" of
"Optional[str]" has no attribute "lower"

conservancy_beancount/reports/rewrite.py:563: error: Argument 2 to
"_iter_yaml" of "RewriteRuleset" has incompatible type "Union[Any, str,
None]"; expected "str"

tests: Replace skipUnless with skipif.

The function skipUnless is available in `unittest`, but not `pytest`.

reconcile: Update helper to show N/A when no results, similar to Perl.

Also clarified that a full account name should be passed. This aligns with the
Perl behaviour.

reconcile: Convert Perl reconciliation helper to Python.

This is a Python version of bc-reconcile-helper.plx.

`expense-type` auto assignment — Expenses:Occupancy

"Expenses:Occupancy" is rarely a program expense, and is almost
always management expense, even when done for a project.
Automatically assign it as such.

Expenses:Payroll:Benefits:Other — `payroll-type` of `US:403b:Other`

Certain adjustments to 403(b) accounts that benefit the employee
should be classified with this account and the `payroll_type` added
herein.

meta_payroll_type — CA:PTO — distinguish between Earned and Taken

Metadata `payroll-type` will go on `Liabilities:Payable:Vacation` and
`Expenses:Payroll:Salary`, but we need to distinguish as to whether
the Payroll was earned or taken.

Implement the `CA:` version of the `US:` ones already done.

meta_payroll_type — US:PTO — distinguish between Earned and Taken

Metadata `payroll-type` will go on `Liabilities:Payable:Vacation` and
`Expenses:Payroll:Salary`, but we need to distinguish as to whether
the Payroll was earned or taken.

Later, `Liabilities:Payable:Vacation` should require a
`payroll-type` metadata of either `US:PTO:Taken` or `US:PTO:Earned`,
and the `CA:` equivalents.

query: Let the user interrupt interactive queries.

This makes it easier to iterate on a query because you don't have to restart
the program and reload the books if something goes sideways.

query: Refactor DBColumn.

Avoid an issubclass check on every call, and make it easier for subclasses
to override part of the call implementation.

reports: BaseODS puts each line of strings in a P tag.

This seems to be the most straightforward way to get Calc to automatically
determine a nice row height for multi-line string cells. This has become a
lot more noticeable now that query-report supports putting postal addresses
in cells.

query: Clean unused code.

This moved into _EnvironmentMixin.with_config().

query: set() updates from most Iterables, not just Sequences.

This lets it work correctly on set columns.

query: Add ContextMeta function.

Deduplicate metadata lookup code.

query: Convert query functions that return List to Set.

Beancount's built-in renderers expect this and are better equipped for it.

rtutil: Add RTDateTime class.

See comments for rationale.

config: Add RTCredentials.idstr() method.

Want to reuse this code for a query-report cache key.

query: Add original query as a comment to ODS output.

This is generally helpful to keep for future reference, plus this
presentation can support multiple queries and is more discoverable than the
previous file property.

query: Add a user hint for query TypeErrors.

The error message that native bean-query provides when this happens
confuses many users. Add a hint to point them in the right direction.

query: Development cleanup.

Reorder classes for nicer readability. Put shorter classes higher up, keep
related classes together.

Add developer-facing comments.

Remove unused imports.

query: Improve formatting of ODS output.

* Provide dedicated formatting for more Beancount types.
* Improve code to determine when we're looking up link metadata
and should format output as links.

query: Add BQL functions for dealing with link metadata.

query-report was heading to a place where it was going to bifurcate. You
could structure input with its own special input switches, and with ODS
output, it would have its own dedicated grouping logic and use that.

But those things shouldn't be tied together for users. Instead, add
functions to BQL to be able to do the kind of grouping we want. This commit
adds those. Next we'll extend the ODS output to detect and format these
groups correctly.

query: Add support for --numberify.

bean-query compatibility.