Free more disk space in CI workflows to fix minikube storage failures

[lirenjie95]

Problem

CI jobs that run minikube plus a full image build keep flaking on disk space:

X Exiting due to RSRC_DOCKER_STORAGE: Docker is out of disk space! (/var is at 99% of capacity)

• integration test (04.workflow-false) on main, 2026-07-23 (original attempt; the run was later re-run and passed): https://github.com/eclipse-symphony/symphony/actions/runs/30039594967/attempts/1
• suite build job, same error: https://github.com/eclipse-symphony/symphony/actions/runs/26763226111

minikube start pulls the kicbase image, mage build:all builds all images, and mage cluster:up loads them into minikube — the images end up stored twice under /var/lib/docker. These jobs pass sometimes and flake other times, so the disk headroom is right at the edge.

Fix

The Free up disk space step is identical in 5 workflows (integration.yml, go.yml, fault.yml, suite.yml, release.yaml). suite.yml and fault.yml have the same minikube + full image build profile as integration.yml and will hit the same wall eventually; release.yaml builds multi-arch images where a failure becomes a release incident. This PR extends all five copies uniformly, following the standard list from https://github.com/jlumbroso/free-disk-space while keeping the existing hand-written rm style (no external action):

• /opt/ghc (~5 GB), /opt/hostedtoolcache/CodeQL (~2-3 GB)
• /usr/lib/jvm, /opt/microsoft/powershell, /usr/local/lib/node_modules, /usr/local/.ghcup, /opt/hostedtoolcache/Ruby (~0.5-1 GB each)

df -h / is printed before and after so the freed amount is visible in the logs. /opt/hostedtoolcache is not removed wholesale since setup-go installs Go there. No workflow in this repo uses CodeQL, setup-java, setup-node, or maven/gradle, so these deletions are safe everywhere.

Verification

The before/after df output from this PR's integration 04.workflow-false job is posted in a comment below. release.yaml only runs on release so it cannot be exercised here, but its change is identical to the other four files.

[lirenjie95]

Evidence from the test (04.workflow-false) job of this PR — df -h / before and after the extended Free up disk space step:

Before

Filesystem      Size  Used Avail Use% Mounted on
/dev/root       145G   59G   87G  41% /

After

Filesystem      Size  Used Avail Use% Mounted on
/dev/root       145G   32G  113G  22% /

~27 GB freed (previously the same step freed only ~10 GB), giving minikube + full image builds comfortable headroom. The job passed.

https://github.com/eclipse-symphony/symphony/actions/runs/30279340656/job/90021422571