daily-fix: fix red main: test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints — call explore_persona_space.orchestrate.e

Overview / Motivation

Auto-filed by the watcher urgent-park router (#1681, autonomous_session_watch.urgent_wf_park_pass) from an URGENT (urgency: main-red) parked workflow-fix candidate raised on task #1791. The named test is red on origin/main NOW — every intervening session's Step 9c gate re-classifies it until this fix lands.

Goal

fix tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints red on origin/main: call explore_persona_space.orchestrate.env.load_dotenv() before the heavy import in scripts/issue1773_analyzer_plots.py per #847.

Workflow gap

• Bug observed (emitter's claim, candidate block): tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints fails on pristine origin/main — scripts/issue1773_analyzer_plots.py has a module-top heavy import at line 25 with no load_dotenv() call (verified by the #1791 Step 9c gate run + pristine-scratch oracle strip, 2026-07-29).
• Failing node (router-verified): tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints
• Confidence (emitter): high
• verified-at-filing: uv run pytest tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints -q -> rc=1 at main @ cfde049a3c (2026-07-29T13:23:30Z) (FAILED — red confirmed)

Proposed change (candidate diff sketch — refine in planning)

(see the verbatim candidate block under ## Provenance — the router forwards it unmodified and never synthesizes fields)

Scope / surfaces

• Primary target: scripts/issue1773_analyzer_plots.py
• Failing node: tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints

Constraints / invariants

• NON-workflow-surface fix (wf_fix: false declared by the parking session — the /daily route-2 analogue); scope stays on the named target, never tasks/ state.

Provenance

• fingerprint: 67e489c5ed68
• routed-by: autonomous_session_watch urgent-wf-park-router (#1681)

target_file: scripts/issue1773_analyzer_plots.py bug_observed: tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints fails on pristine origin/main — scripts/issue1773_analyzer_plots.py has a module-top heavy import at line 25 with no load_dotenv() call (verified by the #1791 Step 9c gate run + pristine-scratch oracle strip, 2026-07-29). why_workflow_gap: a live-red workflow-surface scan pin forces every in-flight session's Step 9c gate / Step 10d TG legs to re-classify the red until fixed (the #1643 fleet-wide per-hour cost shape). NOTE: the offending file itself is EXPERIMENT-script scope (wf_fix: false) — route as a daily-fix/implementer task, not a wf-fix session. proposed_change: call explore_persona_space.orchestrate.env.load_dotenv() before the heavy import in scripts/issue1773_analyzer_plots.py per #847. diff_sketch: |

• from explore_persona_space.orchestrate.env import load_dotenv
• load_dotenv() <before the module-top heavy import at line 25> confidence: high urgency: main-red failing_test: tests/test_shared_vm_thread_caps.py::test_no_new_torch_before_dotenv_vm_entrypoints wf_fix: false related_task: #1791