gallivm: Run nir_convert_to_lcssa before nir_convert_from_ssa

gallivm: Run nir_convert_to_lcssa before nir_convert_from_ssa

Without loop exit phis, the code emitted by nir_lower_non_uniform_access
won't be lowered to registers. Therefore, all lanes will have the value
of the last iteration.

Fixes the remaining descriptor indexing fails:
dEQP-VK.descriptor_indexing.storage_texel_buffer
dEQP-VK.descriptor_indexing.storage_texel_buffer_after_bind
dEQP-VK.descriptor_indexing.storage_texel_buffer_after_bind_in_loop
dEQP-VK.descriptor_indexing.storage_texel_buffer_after_bind_in_loop_lifetime
dEQP-VK.descriptor_indexing.storage_texel_buffer_after_bind_lifetime
dEQP-VK.descriptor_indexing.storage_texel_buffer_in_loop
dEQP-VK.descriptor_indexing.storage_texel_buffer_in_loop_lifetime
dEQP-VK.descriptor_indexing.storage_texel_buffer_lifetime
dEQP-VK.descriptor_indexing.storage_texel_buffer_minNonUniform

cc: mesa-stable

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24661>