feat(vllm-tensorizer): Update vllm-tensorizer cloned repository, build with vllm-flash-attn, other optimizations

[sangstar]

vllm-tensorizer hasn't had updates since vLLM's formal adoption of tensorizer model loading. An update to build for the most recent commit to vLLM that includes sharded tensorizer support is presented, along with some fixes to successfully build vLLM with recent updates to the source code. These include:

• Building vLLM's wheel from vLLM's source code proper, rather than CoreWeave's vLLM fork (reflecting its official adoption of tensorizer)
• vLLM's adoption of cmake
• Updated xformers version to 0.0.26.post1
• vLLM now formally using their own forked version of flash-attn, which is built here from source

[github-actions]

@Eta0 Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/13316885649
Image: ``

[github-actions]

@Eta0 Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/13318197054
Image: ``

[github-actions]

@Eta0 Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/13319222913
Image: ``

[github-actions]

@sangstar Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/13397061488
Image: ``

[github-actions]

@sangstar Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/14085967310
Image: ``

[github-actions]

@sangstar Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/14864987607
Image: ``

[github-actions]

@sangstar Build complete, success: https://github.com/coreweave/ml-containers/actions/runs/14890901257
Image: ``