ENH: Add support for np.array literals

[timothydijamco]

This PR allows np.array literals in the PySpark backend. This was already supported on the Pandas backend.

import numpy as np

from ibis import backends
from ibis.expr.types import literal

spark = ...

ibis_pyspark = backends.pyspark.connect(spark)

ibis_pyspark.execute(literal(np.array([1, 2])))

Before:

AttributeError: 'numpy.int64' object has no attribute '_get_object_id'

After:

array([1, 2], dtype=int32)

[jreback]

can you add a release note as well. ping on green.

[timothydijamco]

#2605 will take care of one of the failing tests (SQLite is supposed to skip test_np_array). I'll have to make a few more changes to fix the rest

[timothydijamco]

All tests are passing now. I increased the scope slightly—this PR adds support for (1-d) np.array as literals for backends that support array/list literals in general (basically, all backends excluding MySQL and SQLite)

[jreback]

thanks @timothydijamco