Bump kelos-workers and kelos-pr-responder task memory to 1Gi
What type of PR is this?
/kind cleanup
What this PR does / why we need it:
Bumps the claude-code container memory request from 512Mi to 1Gi for the kelos-workers and kelos-pr-responder TaskSpawners.
Multiple recent kelos-workers-issue-comment-* tasks failed with pod evictions on the node. Inspecting the evicted pods showed claude-code routinely using ~1.2–1.3 GiB of memory while only requesting 512 MiB:
The node was low on resource: memory. Threshold quantity: 100Mi, available: 75424Ki.
Container claude-code was using 1360064Ki, request is 512Mi, has larger consumption of memory.
Because the pods are Burstable QoS, they're first in line for kubelet eviction under node memory pressure. Bumping the request to 1 GiB brings it closer to actual usage and reduces the eviction risk. Scope is limited to the two spawners running the heaviest claude-code workloads — the other spawners can be revisited if they start showing the same failure mode.
Which issue(s) this PR is related to:
N/A
Special notes for your reviewer:
Only requests.memory is bumped; no limit is set on memory (existing behavior preserved). ephemeral-storage and cpu are unchanged.
Does this PR introduce a user-facing change?
NONE
Summary by cubic
Increase claude-code memory request from 512Mi to 1Gi for kelos-workers and kelos-pr-responder to prevent pod evictions under memory pressure. Only requests.memory changed; CPU and ephemeral-storage are unchanged, and there is still no memory limit.