fix: enable schedule commands against OSS Conductor

[v1r3n]

Summary

• The schedule subcommands all bailed out with "Not supported in OSS Conductor" if --server-type wasn't Enterprise. But conductor-oss/conductor ships a scheduler/ module (upstreamed by Orkes) mounted at /api/scheduler/* — exactly the paths the conductor-go SDK already calls. The guards were stale and locked OSS users out of a feature that works fine against the default OSS server.
• Removes the 7 OSS guards, fixes a duplicate-detection bug that hit every OSS schedule create, fixes schedule search, and improves 404 messaging when the scheduler module isn't loaded.

What changed

cmd/scheduler.go:

• Drop isEnterpriseServer() early-returns from list / get / create / update / delete / pause / resume / search.
• Add parseSchedulerAPIError: on a 404 from a scheduler endpoint, return a clear hint that the scheduler module may not be enabled, plus a curl verification command.
• Bug fix — duplicate detection on OSS. OSS Conductor returns 200 OK with an empty body when a schedule doesn't exist (Orkes returns 404). The SDK's GetSchedule unmarshals that into a zero WorkflowSchedule{} and does not error, so the pre-save existence check thought every name already existed. Now also requires existing.Name != "" to treat the schedule as present. Handles both servers uniformly.
• Bug fix — schedule search. It required a positional arg, then iterated args issuing the same query each time. Now runs once with --workflow / --status / --count flags. The --workflow flag was being read by the handler but never registered — now declared on searchSchedulerCmd.
• Cleanup: removed an empty if update {} block; surface non-404 errors from the pre-save GetSchedule instead of silently treating them as "doesn't exist".

CLAUDE.md, README.md:

• Replace the "Schedule commands are only available with Orkes Conductor (Enterprise)" callouts with a note that schedule commands work against both, provided the OSS server includes the scheduler module (the default jar started by conductor server start does).

Test plan

• go build ./... clean
• go test ./cmd/... passes
• Verified end-to-end against a local OSS server (conductor server start):
  • schedule create -n hello -c "0 0 * ? * *" -w hello_world → succeeds (was failing as "already exists" before the duplicate-detection fix)
  • schedule list → shows the new schedule
  • schedule get hello → returns full JSON
  • schedule update -n hello -c "0 0 12 ? * *" -w hello_world → updates cron
  • schedule update -n nope ... → "no such schedule by name nope"
  • second schedule create -n hello ... → correctly errors as duplicate
  • schedule delete hello -y → "deleted successfully"; subsequent schedule list is empty
• E2E bats suite (test/e2e/schedule.bats) — not run in this PR; needs CI with a live Conductor server
• Smoke against an Orkes Conductor (Enterprise) instance to confirm no regression on the 404 path

🤖 Generated with Claude Code