Extend testing of CL_UNORM_INT_101010_2

[gorazd-sumkovski-arm]

All existing tests in test_image_streams, that are capable of testing image formats using the CL_UNORM_INT_101010_2 data type, now do so.

[CLAassistant]

All committers have signed the CLA.

[bashbaug]

I did a search for CL_UNORM_INT_101010, thinking that these are places that likely need to be updated for CL_UNORM_INT_101010_2. Can you check that these don't need updates?

1. get_32_bit_image_format()
2. compare_scanlines() - this one is probably OK.
3. get_format_max_int()
4. get_format_min_int()
5. read_image_pixel()
6. get_pixel_bytes()

There are a few others in other test suites also, but maybe those are intended for another PR.

[gorazd-sumkovski-arm]

1. get_32_bit_image_format() - Added (although this function isn't used anywhere).
2. compare_scanlines() - As you suggested, no need for a special case because all 32 bits in CL_UNORM_INT_101010_2 are used.
3. get_format_max_int() - Added.
4. get_format_min_int() - Added.
5. read_image_pixel<T>() - Already present in read_image_pixel_float(). Seems like read_image_pixel<T>() is used for int and uint while float has a separate function. As a side note, it looks like the cases for 101010, 555 and 565 in read_image_pixel<T>() are wrong, they are missing the normalisation divisions.
6. get_pixel_bytes() - Added.

[bashbaug]

I'd be surprised if there weren't a few more places that need updates for 101010_2, but I didn't find anymore cases that are obviously missing. Thanks!

[kpet]

Any objections to merging this PR ahead of the next teleconference? It is blocking the addition of testing for the 2_101010 format added by KhronosGroup/OpenCL-Docs#1223.

[bashbaug]

I'm going to go ahead and merge this. Since these are purely additions I think the risk of problems for implementations that do not support the 101010_2 format is low. If we find problems we can always revert it.

@lakshmih FYI.

[kpet]

Thanks!