custatevec for aarch64/ppc64le requires glibc 2.27+

[leofang]

See #47 (comment). Once the conda-forge infrastructure is ready, we should restore these two lines

cuquantum-feedstock/recipe/meta.yaml

Lines 99 to 102 in ce8944d

	# this should be set once conda-forge supports newer glibc
	# https://github.com/conda-forge/conda-forge.github.io/issues/1941
	# run_constrained:
	# - __glibc >=2.27 # [aarch64 or ppc64le]

Right now they are commented out because we need to consume them from within the CI (to build cuquantum-python).

[leofang]

FYI. We're not alone. xref:

• cuDNN v8.8.0 for aarch64 + CUDA 12 is not consumable by recipe maintainers cudnn-feedstock#58
• libcublas on aarch64 requires glibc 2.27 libcublas-feedstock#13
• cuSPARSELt v0.4.0 for aarch64 requires glibc 2.27 cusparselt-feedstock#29

[leofang]

Note: When the infra is ready, we should switch this line

cuquantum-feedstock/conda-forge.yml

Line 20 in d21a50c

test: native

to use native_and_emulated to turn on QEMU at the test stage (xref: #47 (comment)).