Add complex Shopify template rendering tests

[benfaerber]

Summary

Adds end-to-end rendering tests built from 6 real-world templates taken from Shopify's own Ruby Liquid benchmark suite (performance/tests/{tribble,vogue,dropify,ripen}). These give the renderer a comprehensive, real-world workout beyond the existing focused unit tests.

Templates added under test/templates/complex/:

File	Source theme	Exercises
theme.liquid	vogue	layout, menus, blog list (limit/offset), featured products, pluralize, date
minicart_theme.liquid	dropify	minicart loop, pluralize, money, linklists
index.liquid	tribble	product grid, variant loops, price logic
product.liquid	tribble	tags loop, variant <select>, image loops
product_variants.liquid	ripen	options, vendor/type, price_varies
cart.liquid	tribble	nested loops, cycle, money_with_currency, forloop.first, limit

How it works

• A single shared, representative data context (shop, cart, product, collections, linklists, pages, blogs, …) is built with small helper constructors.
• A universal passthrough filter stub covers the Shopify-specific filters liquid-ml doesn't implement (asset_url, product_img_url, link_to, within, stylesheet_tag, …). The standard library is consulted first, so real filters (money, pluralize, truncate, …) keep their behavior.
• Each template is rendered and asserted to be non-empty and contain stable marker substrings.

Notes / findings

• The renderer handled all 6 templates end-to-end: nested for-loops, cycle, forloop.first/last, hyphenated keys (linklists.main-menu), list .size/.first, and nested conditionals.
• One correctness behavior is encoded in the assertions: {% for article in blogs.news.articles limit: 6 offset: 1 %} correctly skips the first article, so the test asserts News Two rather than News One.

Testing

dune test — 268 tests run, all passing (was 262; +6).

[codecov]

Codecov Report

❌ Patch coverage is 95.94595% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
test/test_complex_templates.ml	95.94%	3 Missing ⚠️

📢 Thoughts on this report? Let us know!