Work split out from #633

- Reorgs the unit test fixtures for reuse by mlos_viz (future PR)
- Removes TunablesGroups from Storage classes. Currently unnecessary,
belongs in Experiment, and causes inconsistencies with ExperimentData
interactions.
- Adjusts the initialization code to match.

---------

Co-authored-by: Sergiy Matusevych <[email protected]>

…rty for `TrialData` and `ExperimentData` (#648)

Useful for grouping trials by the config they used. In use by upcoming
#633 for generating graphs with variance error bars for repeated
configs.

Also refactors a number of other things:
- Standardize on `experiment_id` instead of `exp_id`
  (but not the db schema for now)
- Standardize on `tunable_config_id` instead of `config_id` in the API
since we also call it `tunable_config` for the object fetching property
to distinguish from the `config` dict used internally.
  (but not the db schema for now)
- Rework the idea of `TunableConfigTrialGroup` as an object inaddition
to an ID (further methods can be added later to move back and forth
between types when doing interactive analysis).
- Rework the idea of a `TunableConfig` as an object for fetching tunable
value assignments (similar justification - easier grouping in the future
by fetching trial across experiments based on config - eventually could
be used to house the experiment merge logic).
- Rename `results` APIs to `results_df` (similar for others that return
`pandas.DataFrame`) to match the `results_dict` that return `dict`
- Refactor test fixtures to match other styles and for future use (moved
to #644).
- Expand tests

NOTE:
- We cut a new version with this commit since there are potentially
breaking API changes (e.g., `results` -> to `results_df` and `exp_id` ->
`experiment_id`).

Currently builds off of #644 and splits work out of #633

---------

Co-authored-by: Sergiy Matusevych <[email protected]>

Splitting work out from #633.

Having an easy way to identify what is the "default" config group for an
Experiment makes analysis of improvements easier.

One challenge with this is that we don't currently strictly enforce
non-changes between resumptions of an Experiment, so it's currently
possible for the user to change the tunables config and adjust what the
"default".

Moreover, in the past we didn't store what the default config was in the
DB.

By convention we take the first one for now as a fallback.

#628 mistakenly included an early attempt at adding
`optimization_target` and `optimization_direction` to the `experiment`
table in the `mlos_bench.storage.sql` backend.

In that PR we later moved it to its own `objectives` table to eventually
support multi-objectives.

Nothing accesses those columns now, however including them in the
metadata makes it impossible to load storage backends previously created
with the old schema since adjusting columns with sqlalchemy's
`create_all()` API only considers table existence.

On the contrary, the latter means that we will automatically support old
storage backends with the new code for the `objectives` table.

Removing these two columns in the `metadata` schema description simply
allows that to proceed without error.

See Also: #649

Co-authored-by: Sergiy Matusevych <[email protected]>

…` is given. (#651)

Splitting work out from #633.
Bug fix for #642.

Co-authored-by: Sergiy Matusevych <[email protected]>

Polish for a demo.

See Also: https://stackoverflow.com/a/61185987

---------

Co-authored-by: Sergiy Matusevych <[email protected]>

- Mark `mlos_viz` as `typed` for `mypy`
- Bump version
- Mock calls to matplotlib/dabl for testing
- Add plotting of top-N configs
- Improve plots for handling repeat config trials via variance error
bars

---------

Co-authored-by: Sergiy Matusevych <[email protected]>

…res plugin (#656)

See Also:
- TvoroG/pytest-lazy-fixture#65
- pytest-dev/pytest#11890