Refactor the data publication pipeline for KG to the cloud pipeline

[matentzn]

Description of the changes

This PR is migrating the KG HF pipeline to the cloud environment and so that the tagging system is fully supported.

Fixes / Resolves the following issues:

• https://linear.app/everycure/issue/XDATA-284

Checklist:

• Added label to PR (e.g. enhancement or bug)
• Ensured the PR is named descriptively. FYI: This name is used as part of our changelog & release notes.
• Looked at the diff on github to make sure no unwanted files have been committed.
• Made corresponding changes to the documentation
• Added tests that prove my fix is effective or that my feature works
• Any dependent changes have been merged and published in downstream modules
• If breaking changes occur or you need everyone to run a command locally after
  pulling in latest main, uncomment the below "Merge Notification" section and
  describe steps necessary for people
• Ran on sample data using kedro run -e sample -p test_sample (see sample environment guide)

I tested this, but note that the tag that we ended up USING was v0.13.0 for some reason:

#!/bin/bash

source .env

if [ -z "$HF_TOKEN" ]; then
  echo "Error: HF_TOKEN environment variable is required"
  exit 1
fi

RELEASE_VERSION=0.15.0

# GCS auth for Spark Hadoop connector (locally uses ADC instead of metadata server)
export GOOGLE_APPLICATION_CREDENTIALS="${GOOGLE_APPLICATION_CREDENTIALS:-$HOME/.config/gcloud/application_default_credentials.json}"

# Required by cloud globals.yml but not used by data_publication
export RUN_NAME="${RUN_NAME:-unused}"
export RUNTIME_GCP_PROJECT_ID="${RUNTIME_GCP_PROJECT_ID:-unused}"
export RUNTIME_GCP_BUCKET="${RUNTIME_GCP_BUCKET:-mtrx-us-central1-hub-dev-storage}"

uv run kedro run -e cloud --pipeline=data_publication