… `__getitem__` (#1082)

* feat: add warning when accessing structs with wrong method

* polishing words

* move test from unit to system

* 🦉 Updates from OwlBot post-processor

See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md

* add more tests

* polish tests

* upgrade test setup

* define a dedicated warning and set a stack level at warning site

* fix format

* 🦉 Updates from OwlBot post-processor

See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md

* fix doc

* Apply suggestions from code review

* update doc

---------

Co-authored-by: Owl Bot <gcf-owl-bot[bot]@users.noreply.github.com>
Co-authored-by: Tim Sweña (Swast) <swast@google.com>

* feat: support context manager for bigframes session

* add table cleanup tests for session context manager

* rename ctor, split large test

… models (#1096)

* feat: allow `fit` to take additional eval data

The additional eval data would be used to measure the fitted model and
attach the measurement to the underlying BQML model, which can be used
as benchmark for the model consumers in BigQuery Studio and otherwise.

* subclass from TrainablePredictor

* add support for fit-time evaluation in ensemble models

* fetch logistic regression eval numbers from multiClassClassificationMetrics

* use the generic type template

* update vendored docstrings for fit taking X_eval, y_eval

* update key to fetch model eval metrics

* enfore binary classification in the logistic regression test

#1108)

* perf: improve series.unique performance and replace drop_duplicates in tpch test.

* change variable name

* update docstring and index.

* update flag to control behavior.

* update examples

* delete empty line

* keep order error

* reduce query_count

* improve q16

* benchmark updates

…tor predict (#1119)

* feat: add  option for GeminiTextGenerator predict

* add pricing link to the warning message

Co-authored-by: release-please[bot] <55107282+release-please[bot]@users.noreply.github.com>