Add theory application pages (conflicts, semantic actions, full example)

[halotukozak]

Summary

• theory/conflicts.md — Conflicts & Disambiguation: formal parse table conflict definition (s, t, Shift/Reduce), shift/reduce conflict with 1+2+3 example and real Alpaca error message, alwaysBefore/alwaysAfter discrepancy callout, reduce/reduce conflict, before/after resolution DSL bridge
• theory/semantic-actions.md — Semantic Actions & Typed Results: S-attributed translation scheme, formal definition f(v₁,...,vₙ) → vₐ, maps { $$ = $1 + $3 } notation to Alpaca's case (Expr(a), PLUS(_), Expr(b)) => a + b, full 7-production CalcParser table, no-parse-tree explanation (actions fire immediately on reduce)
• theory/full-example.md — Full Calculator: Theory to Code (185 lines): 6-step narrative from bare grammar → ShiftReduceConflict compile error → resolved CalcParser → 1+2*3=7.0, semantic action trace, theory-to-code mapping table cross-linking all 8 prior theory pages

Part of the v1.1 Compiler Theory Tutorial milestone — Phase 10: Application (TH-07, TH-08, TH-09).

Test plan

• ./mill docJar passes (all sc:nocompile blocks correctly annotated)
• theory/conflicts.md has formal conflict definition and alwaysBefore/alwaysAfter note
• theory/semantic-actions.md says "S-attributed", no "inherited attribute" / Rule[Int]
• theory/full-example.md uses production.div (not production.divide), Rule[Double], shows 7.0 result
• All pages have > **Compile-time processing:** callout and cross-links to reference docs

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR adds three comprehensive theory documentation pages that bridge compiler theory concepts with Alpaca's practical implementation. The pages complete Phase 10 of the v1.1 Compiler Theory Tutorial milestone by demonstrating how conflicts, semantic actions, and the full calculator example work together.

Changes:

• Adds formal treatment of parse table conflicts (shift/reduce and reduce/reduce) with error message examples and resolution strategies
• Adds semantic actions documentation explaining S-attributed translation, extractor patterns, and typed results without exposing parse trees
• Adds complete end-to-end calculator example showing the progression from bare grammar to resolved parser with execution trace

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
docs/_docs/theory/conflicts.md	Defines parse table conflicts formally, explains shift/reduce and reduce/reduce conflicts, documents the before/after resolution DSL, and notes the alwaysBefore/alwaysAfter error message discrepancy
docs/_docs/theory/semantic-actions.md	Explains S-attributed translation scheme, maps formal semantic action notation to Alpaca's case syntax, provides complete production action table, explains immediate evaluation without parse tree construction
docs/_docs/theory/full-example.md	Walks through 6-step calculator implementation from lexer to running code with 7.0 result, includes semantic action trace, theory-to-code mapping table, and compile-time processing explanation

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The terminal symbol should be "NUMBER" (all caps) not "Number" to match the terminal naming convention consistently used throughout the documentation. The example productions should be "Integer → NUMBER" and "Float → NUMBER", and the error message should display "Reduce Integer -> NUMBER vs Reduce Float -> NUMBER".

Suggested change

	Reduce Integer -> Number vs Reduce Float -> Number
	In situation like:
	Number ...
	Reduce Integer -> NUMBER vs Reduce Float -> NUMBER
	In situation like:
	NUMBER ...

[Copilot]

This line references "STATE.md", which appears to be an internal project management file that is not part of the user-facing documentation. This reference should be removed as users won't have access to this file and it disrupts the flow of the documentation.

Suggested change

	Decision confirmed: "Parse tree never exposed in Alpaca — semantic actions evaluated immediately during LR(1) reduce; parse() returns typed value directly." (STATE.md)
	This design is intentional: the parse tree is never exposed in Alpaca — semantic actions are evaluated immediately during LR(1) reduce, and `parse()` returns the typed value directly.