task: #3625354 Update @vardot/varbase-e2e to 2.0.6

Issue: #3625354

Problem/Motivation

The suite pinned @vardot/varbase-e2e at the open ^2. Every other Vardot functional suite (varbase_starter, rightup, varbase_core, varbase_editor, varbase_search and ~13 more) pins the caret to the exact version it was verified on, so ^2 was the outlier here.

2.0.6 fixes three things that hit this repo directly, since the pipeline runs 18 parallel functional buckets of which five drive the Drupal Canvas editor:

• #53 the Canvas editor open step timed out under load before its own retry loop could report.
• #58 each smartSettle stage now gets its own share of the budget.
• #57 smartSettle could wait forever: remaining() floored to 0, and Playwright reads a timeout of 0 as no timeout at all.

Proposed resolution

Tighten the constraint to ^2.0.6, the tested release. package-lock.json is gitignored, so the package.json constraint is the only thing that pins what CI resolves. This raises the floor to the verified version while still allowing patch and minor updates inside 2.x.

Verification, full suite run against a live Horizon Aid install

Both runs were made on the same clean checkout of 1.0.x against the same live site (plain drupal/cms plus Horizon Aid 1.0.x, 70 nodes, 9 Canvas pages, 74 media, CI's test preparation applied), same command, --parallel 2, --tags "not @wip", cucumber retry: 1. The only variable was the varbase-e2e version.

	before (2.0.4)	after (2.0.6)
Scenarios	367 (10 failed, 357 passed)	367 (3 failed, 364 passed)
Steps	3108 (8 failed, 17 skipped)	3108 (3 failed, 38 skipped)
Wall time	93m 28s	33m 24s

All five Drupal Canvas editor failures are gone. Every one of them was function timed out ... within 180000 milliseconds on "opens in the Canvas editor", exactly the #53 failure mode:

• 10-02 Home / About Canvas pages
• 10-03 Our Programs Canvas page
• 10-04 Resources / Events Canvas pages

Per-bucket step-execution time (minutes), before to after:

• 10-drupal-canvas 21.5 to 5.7 (0 failures, was 5)
• 07-quality 28.2 to 5.6 (0 failures, was 2, both 45s step timeouts)
• 09-editorial 57.3 to 31.8
• 11-permissions 10.3 to 6.1
• 12-search 1.7 to 1.5 (0 failures in both runs)
• total 148.9 to 65.6

Remaining 3 failures, none caused by this change:

• 04-01-resources-listing.feature (2). Fails identically at 2.0.4 and at 2.0.6, and reproduces in isolation. The "Community-led reconstruction" node exists, is published and serves 200, but on this install it sorts onto page 3 of a 12-per-page listing of 26 published blog nodes, so it is not among the Latest Updates the scenario expects. Test-site content ordering, not a code regression.
• 09-06-media-crud.feature (1). A page.waitForFunction 10s timeout under 2-worker contention on a memory-constrained machine. Passes 3/3 scenarios, 148/148 steps when re-run in isolation. A load flake, not a regression.

Remaining tasks

• Review by a human.

User interface changes

None.

API changes

None.

Data model changes

None.

AI-Generated: Yes

Checkpoints:

• File an issue
• Addition/Change/Update/Fix
• Testing to ensure no regression
• Automated unit testing coverage
• Automated functional testing coverage
• UX/UI designer responsibilities
• Readability
• Accessibility
• Performance
• Security
• Developer Documentation
• User Guide Documentation
• Reviewed by human
• Code review by maintainers
• Full testing and approval
• Credit contributors
• Review with the product owner
• Release notes snippet
• Release