perf(engine): skip the redundant render-pass re-parse of define-free templates

[alexmond]

Summary

jhelm parses every template twice per render: a collect pass (collectNamedTemplates) parses each template into the factory under its helm-style key to gather define blocks, then the render pass (renderChartTemplates) re-parses the same (name, text) before executing it. Because parseWithCache's key includes the name, even a wired TemplateCache can't dedupe the two passes within a single render — and template parsing is jhelm's single hottest render path (profiled below).

This adds a small per-render parsedTextHash memo: when the identical text is already parsed under a name and it declares no define block, the render-pass re-parse is skipped (it would only re-create the identical root node).

Why it's safe

• Templates that declare {{ define }} — which could re-order the global define table — still parse, so define precedence is untouched.
• Key collisions (distinct text under the same helm-style key) still parse and overwrite, so collision resolution is untouched.
• The text.contains("define") check is a cheap superset guard: a false positive only costs a redundant parse, never correctness.
• The memo is reset each render in doRender alongside the factory; the factory.getRootNodes().containsKey(name) guard makes it robust to the per-render factory swap.

Measured impact

60-chart render-only batch (jvmlens, live attach), before → after:

Metric	Before	After	Δ
parseWithCache CPU	42%	31%	−28% samples
parseWithCache allocation	2.3 GB	1.3 GB	−45%
Engine.* total allocation	3.6 GB	3.0 GB	−18%
GC pause	448 ms	358 ms	−20%

(Other frames' shares rise in the diff purely from share-inversion + more batch rounds completing in the same window — their absolute per-render work is unchanged.)

Verification

• Full clean install green.
• 540/540 parity charts byte-for-byte identical to helm template (0 failures), including the define-precedence-sensitive umbrella charts: gitlab/openbao (Chart signing and verification #21), druid, redpanda/console, istiod.

🤖 Generated with Claude Code

[github-actions]

Overall Project	82.27%	🍏
Files changed	100%	🍏

Module	Coverage
jhelm-core	84.43%	🍏

Files

Module	File	Coverage
jhelm-core	Engine.java	90.96%	🍏

[codecov]

Codecov Report

❌ Patch coverage is 75.00000% with 2 lines in your changes missing coverage. Please review.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...n/java/org/alexmond/jhelm/core/service/Engine.java	75.00%	0 Missing and 2 partials ⚠️

📢 Thoughts on this report? Let us know!

[github-actions]

Overall Project	82.27%	🍏
Files changed	100%	🍏

Module	Coverage
jhelm-core	84.43%	🍏

Files

Module	File	Coverage
jhelm-core	Engine.java	90.96%	🍏