fix(executor): remove the env keys a runFlow or retry added when it ends

[bulatgaleev]

Summary

Maestro gives a sub-flow its own env scope: enterEnvScope saves the env and leaveEnvScope puts that copy back, so a key the sub-flow added is gone when it returns (GraalJsEngine.kt, around runSubFlow in Orchestra.kt, which repeat, retry and runFlow all go through, at cli-2.10.0). maestro-runner's withEnvVars put each key back to the value it had before, and a key that had none was set to an empty string instead of being removed. After a runFlow, retry or sub-flow with env: {KEY: ...}, KEY stayed defined: typeof KEY was "string", $KEY expanded to nothing, and runShell saw KEY="" in its environment. withEnvVars now uses applyScopedEnv, which runScript's env already used (#109): a key is restored when it existed and removed when it did not.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update
• Refactoring (no functional changes)

Changes Made

• withEnvVars, which runFlow, retry and a sub-flow file's own env go through, returns the restore function of applyScopedEnv.
• pkg/executor/scoped_env_test.go: a key the block added is removed and a key it changed gets its old value back, and a runFlow whose env sets SCOPED leaves typeof SCOPED as "undefined" for the next step. Both tests fail without the change.
• A CHANGELOG.md entry.

Related Issues

No existing issue found. #109 (closed) made runScript's env scoped in the same way.

Testing

• go test ./pkg/executor/
• Added tests for new functionality
• Part of our fork's build, which runs a 47-flow suite every night on a physical iPhone (iOS 27, WebDriverAgent reached through a forwarded port). The tests above are the evidence for this change, not that suite
• go test -race for pkg/executor, pkg/driver/wda, pkg/jsengine, pkg/flow and pkg/driver/uiautomator2 passes at the top of the stack; go vet ./..., gofmt -l and golangci-lint v2.13.2 (the version CI pins) report nothing on every branch of the stack
• make test as a whole: not run, because its device tests drive whatever device is attached to the machine
• make lint: the Makefile has no lint target, and the other linters make check runs (staticcheck, revive, errcheck, nilaway, gosec) are not installed here

Checklist

• Code follows project style guidelines
• Self-reviewed the code
• Added/updated documentation as needed (no documentation change needed)
• No breaking changes (or documented if breaking)
• CHANGELOG.md updated (for notable changes)

Additional Notes

After a runFlow with env:, a name the block added is undefined again, so typeof KEY is "undefined" and runShell no longer gets KEY. ${KEY} still expands to nothing.

Stack 3 of 7: #200 → #194 → #195 → #196 → #197 → #198 → #199. Merge in that order. This branch is built on #200 and #194, so it also contains their commits. This PR's own change is the top commit, df20b1b. Once the PRs below it merge, the rest of the diff disappears, and nothing conflicts. All seven sit on main at d3f738f (the merge of #185).

[codecov]

Codecov Report

❌ Patch coverage is 87.50000% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
pkg/executor/scripting.go	87.50%	0 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!