perf(dot-columns): speed up heavily used .columns

[cpcloud]

Description of changes

This PR changes the Table.columns property to return the underlying
tuple of names from the operation's schema attribute.

Right now, every .columns access incurs the overhead of creating a list,
which remained for backwards compatibility.

I was refactoring selectors in #9917, and noticed some performance differences
that boiled down to accessing this attribute many times in the case of
selectors operating on very wide tables.

The other approach I considered was t continue returning a list, and speed up
the access using functools.cached_property.

While this doesn't break the API, it seemed incredibly wasteful to store two
copies of every table's columns, especially in these slower cases where
tables have a ton of columns.

BREAKING CHANGE: Table.columns is now a tuple instead of a list. Your usage of .columns may require updating if you're depending specifically on .columns being a list.

[cpcloud]

The difference in performance is pretty stark:

---------------------------------------------------------------------------------- benchmark 'test_dot_columns[1000000]': 2 tests ---------------------------------------------------------------------------------
Name (time in ms)                                Min                Max               Mean            StdDev             Median               IQR            Outliers             OPS            Rounds  Iterations
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_dot_columns[1000000] (0001_d552afa)     10.3734 (>1000.0)  14.1146 (>1000.0)  10.6059 (>1000.0)  0.4976 (>1000.0)  10.4641 (>1000.0)  0.1202 (>1000.0)      5;11         94.2872 (0.00)         90           1
test_dot_columns[1000000] (NOW)               0.0003 (1.0)       0.0090 (1.0)       0.0003 (1.0)      0.0001 (1.0)       0.0003 (1.0)      0.0000 (1.0)      946;2493  2,904,766.3008 (1.0)      185529           1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

----------------------------------------------------------------------------------- benchmark 'test_dot_columns[100000]': 2 tests ------------------------------------------------------------------------------------
Name (time in us)                                Min                 Max                Mean             StdDev              Median               IQR             Outliers  OPS (Kops/s)            Rounds  Iterations
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_dot_columns[100000] (0001_d552afa)     197.4140 (853.20)   278.9990 (174.87)   207.2664 (849.73)   13.9447 (>1000.0)  201.0715 (829.57)   7.7540 (>1000.0)    186;200        4.8247 (0.00)       1642           1
test_dot_columns[100000] (NOW)                0.2314 (1.0)        1.5954 (1.0)        0.2439 (1.0)       0.0137 (1.0)        0.2424 (1.0)      0.0057 (1.0)      2999;3160    4,099.6842 (1.0)      193051          21
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------------- benchmark 'test_dot_columns[10000]': 2 tests ---------------------------------------------------------------------------------
Name (time in us)                              Min                Max               Mean            StdDev             Median               IQR             Outliers  OPS (Kops/s)            Rounds  Iterations
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_dot_columns[10000] (0001_d552afa)     19.5370 (61.05)    48.8320 (6.99)     21.2852 (61.88)    1.5586 (25.24)    21.0490 (61.73)    0.4710 (469.09)   2090;3084       46.9810 (0.02)      35120           1
test_dot_columns[10000] (NOW)               0.3200 (1.0)       6.9830 (1.0)       0.3440 (1.0)      0.0617 (1.0)       0.3410 (1.0)      0.0010 (1.0)      601;33246    2,907.2263 (1.0)      106519           1
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean
========================================================================================= 3 passed in 9.38s ==========================================================================================

Sure, we're in microsecond territory, but this property is used quite a bit, and hit pretty heavily in selector processing, which tends to invoke .columns in some form of loop.

[cpcloud]

Another approach is to replace the use of .columns in internals, (perhaps with a private _columns property), which is annoying but also not a breaking change.

The downside there is that with .columns and ._columns, developers can easily choose the wrong one and get bad performance.

[cpcloud]

Moving to draft to avoid merging before 9.4

[cpcloud]

The 3 cloud tests that were failing are now passing.

ibis/backends/snowflake/tests/test_client.py::test_insert PASSED         [ 33%]
ibis/backends/bigquery/tests/system/test_client.py::test_parted_column[date] PASSED [ 66%]
ibis/backends/bigquery/tests/system/test_client.py::test_parted_column[timestamp] PASSED [100%]

========================================================================================= 3 passed in 39.24s =========================================================================================