Use std::overflow_error when output would exceed column size limit

[davidwendt]

Description

Replaces generic cudf::logic_error exception with std::overflow_error where appropriate in libcudf.
Since this changes what is thrown in certain APIs, I think this technically is a breaking change.

Closes #12925

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[bdice]

Do we need to do overflow comparisons with operands as std::size_t always? Should we use a common detail function to do these comparisons, like:

#include <limits>

template<typename T>
bool check_column_size_overflow(T size) {
    return static_cast<std::size_t>(size) <= static_cast<std::size_t>(std::numeric_limits<cudf::size_type>::max());
}

[bdice]

Approving to unblock this because I think it's a net improvement. But I'm still in favor of doing this work with a helper function that makes this logic less repetitive across libcudf.

[bdice]

Note: While looking at the MAX_JOIN_SIZE issue, I think there's some cleanup of constants on the Python side that needs to be written in terms of size_type_dtype.

cudf/python/cudf/cudf/_lib/__init__.py

Lines 43 to 46 in 707a480

	MAX_COLUMN_SIZE = np.iinfo(np.int32).max
	MAX_COLUMN_SIZE_STR = "INT32_MAX"
	MAX_STRING_COLUMN_BYTES = np.iinfo(np.int32).max
	MAX_STRING_COLUMN_BYTES_STR = "INT32_MAX"

I'll file a follow-up there.

[davidwendt]

/merge