fix(keyshare): stop logging confusing mode=development for tasks

[dobby-coder]

Summary

The keyshare tasks command logs mode=development on every startup, which is confusing because — unlike the keyshare/myirma/irma servers — the tasks command has no development/production distinction at all (it defines no --production flag).

This is the conservative fix requested in #413: stop logging the misleading line for tasks, without changing any operational defaults.

Changes

• readConfig now only includes the mode field in the startup log line for commands that actually have a dev/production distinction, i.e. those defining a --production flag (irma server, keyshare server, myirma server). The keyshare tasks command logs no mode.
• Added a small helper commandHasProductionMode(cmd) and a regression test (config_test.go) asserting that tasks reports no mode while the server commands do.
• CHANGELOG entry.

Operational behaviour and production defaults are unchanged for all commands.

Testing

• go build ./yivi/..., go vet ./yivi/cli/irmacli/, gofmt — clean
• go test ./yivi/cli/irmacli/ -run TestCommandHasProductionMode -v — passes

Closes #413

🤖 Generated with Claude Code

[rubenhensen]

/dobby fix conflicts

[dobby-coder]

Got it — I'm on it. I'll take a look at PR #578, resolve the merge conflicts, and push the update to the same PR.

[dobby-coder]

VERDICT: approve

Rules Dobby 2 — sign-off (clean)

Conflict resolution verified clean and non-noop against master; build, go vet, gofmt, and the irmacli tests (including the new TestCommandHasProductionMode regression test) pass.

Rule sweep on the changed Go files found no violations:

• Conventional-commit PR title (fix(keyshare): …) ✓
• Closes #413 closing keyword present ✓
• Regression test committed for the fix ✓
• No Go struct/pointer equality traps introduced ✓
• Merge kept the CHANGELOG union and the diff intact ✓

Review Dobby 2 handed forward no findings. No bug- or style-severity issues to fix. Approving.

[dobby-coder]

Conflicts resolved and verified clean on PR #578 — build, go vet, gofmt, and the irmacli tests (including the new TestCommandHasProductionMode regression test) all pass. My rule sweep found no issues, so I posted a VERDICT: approve review comment (using COMMENT since it's self-authored). It's ready for a human sign-off and merge.