Resolve Pydantic to v2

[dlqqq]

Three packages are responsible for forcing Conda to resolve Pydantic to v1 instead of v2. We need to update both Jupyter AI & Scheduler, but amazon-sagemaker-jupyter-scheduler will also need to be updated to loosen the version constraint on Pydantic:

$ pipdeptree --packages pydantic --reverse --depth 1 -w silence
pydantic==1.10.13
├── amazon-sagemaker-jupyter-scheduler==3.0.2 [requires: pydantic==1.*]
├── confection==0.1.3 [requires: pydantic>=1.7.4,<3.0.0,!=1.8.1,!=1.8]
├── jupyter-ai==2.4.0 [requires: pydantic~=1.0]
├── jupyter-ai-magics==2.4.0 [requires: pydantic~=1.0]
├── jupyter-scheduler==2.3.0 [requires: pydantic~=1.10]
├── langchain==0.0.318 [requires: pydantic>=1,<3]
├── langsmith==0.0.60 [requires: pydantic>=1,<3]
├── openapi-schema-pydantic==1.2.4 [requires: pydantic>=1.8.2]
├── spacy==3.7.2 [requires: pydantic>=1.7.4,<3.0.0,!=1.8.1,!=1.8]
├── thinc==8.2.1 [requires: pydantic>=1.7.4,<3.0.0,!=1.8.1,!=1.8]
└── weasel==0.3.4 [requires: pydantic>=1.7.4,<3.0.0,!=1.8.1,!=1.8]

[claytonparnell]

amazon-sagemaker-jupyter-scheduler dependencies loosened as of https://github.com/conda-forge/amazon-sagemaker-jupyter-scheduler-feedstock/commit/1736ece429e91af95f97bf7b86bab09b7de63f64