Fix: dot-index reattaches to multi-arg call result

[danieljohnmorris]

Summary

Closes ILO-52. spl s d .1 (where both args are bare variable references) previously parsed as Call(spl, [Ref(s), Index(Ref(d), 1)]) — the .1 was stolen by the bare-ident arm of parse_atom_body before the outer call loop could claim it.

Root cause

parse_atom_body's bare-Ident branch unconditionally called parse_field_chain, greedily consuming .N even while inside the arg collection loop of an outer call.

Fix

Added an in_call_args: bool context flag to Parser. Set to true for the duration of every greedy arg loop in parse_call_or_atom and parse_call_arg (save/restore pattern). In parse_atom_body's bare-ident branch, when in_call_args is true, field-chain consumption is suppressed — the outer parse_field_chain call at the end of each loop then attaches the .N to the assembled call result.

Before / After

spl s d .1

Before: Call { function: "spl", args: [Ref("s"), Index { object: Ref("d"), index: 1 }] }
After:  Index { object: Call { function: "spl", args: [Ref("s"), Ref("d")] }, index: 1 }

num spl s d .1

Before: Call { function: "num", args: [Call { function: "spl", args: [Ref("s"), Index { object: Ref("d"), index: 1 }] }] }
After:  Call { function: "num", args: [Index { object: Call { function: "spl", args: [Ref("s"), Ref("d")] }, index: 1 }] }

Verified no-regression cases

• spl "1.2.3" "." .1 — literal last arg (already worked)
• (spl s d).1 — explicit parens (always worked)
• x.1 standalone — bare ident not in call args (still works)
• at xs 0 .1 — literal last arg (already worked)

Changes

• src/parser/mod.rs — in_call_args flag + save/restore in 4 arg loops + suppression in bare-ident atom branch
• tests/parser_call_index.rs — new regression test file covering the broken and no-regression cases
• examples/call-dot-index.ilo — new example demonstrating the fix
• examples/call-result-dot-index.ilo — removed the now-obsolete "heads-up" note about bare-ident last args

🤖 Generated with Claude Code

[codecov]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
218	1	217	0

View the top 1 failed test(s) by shortest run time

ilo::codegen::fmt::tests::idempotent_example_04

Stack Traces | 0.011s run time

thread 'codegen::fmt::tests::idempotent_example_04' (30295) panicked at src/codegen/fmt.rs:716:9:
parse errors in test: [ParseError { code: "ILO-P003", position: 42, span: Span { start: 124, end: 138 }, message: "expected `;`, got text `\"Notification\"`", hint: None }]
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[danieljohnmorris]

Superseded by #572 (already merged — same fix for call-result dot-index reattach)