[Version] Bump version to 0.2.80

[akaashrp]

Changes

• Migrate logit bias application, penalty application, and sampling to WebGPU kernels
• Recompile models in binary-mlc-llm-libs after TVM FFI updates (v0_2_80)
• Add enable_latency_breakdown field to extra_body to enable collection of statistics during sampleTokenFromLogits
• Temporarily remove support for q0f32 models due to correctness issues
• Upgrade versions of devDependencies to resolve issues with outdated packages
• Add .nvmrc and migrate CI to node 24
• Update eslint and jest configs

TVMjs / web-runtime

• Upgrade to version 0.23.0-dev1 to maintain compatibility with TVM FFI updates

TVM and MLC commits

The WASMs that the config now pulls were compiled at the following TVM and MLC-LLM commits (mlc-ai/binary-mlc-llm-libs#158):
TVM: apache/tvm@c8515e1
MLC-LLM: mlc-ai/mlc-llm@4084e7f

[CharlieFRuan]

This is great work. Thank you so much!