docs: add Quick Start and FAQ with JSON-LD schema

[cloudsealed]

Add 60-second Quick Start at the top for devs unfamiliar with LLVM/Numba.

Enrich existing FAQ with: Numba comparison, GPU requirement, Python version.

Add JSON-LD FAQPage schema for LLM/search discoverability (Perplexity, Google SGE).

[github-actions]

💰 cloudsealed-jit — billing waste audit

Analysed 31 day(s) of billing (124 line items) totalling USD 23,990.88. Mean daily spend USD 773.90, stability ratio 1.86. 3 anomalous day(s) detected; 11.9% of total spend sits above the modelled baseline. Identified up to USD 16,341.33 in addressable monthly spend.

Waste	Avg daily cost	Stability	Anomalies
11.9%	$773.90	1.86	3

Anomalies

Date	Severity	z-score	Deviation
2026-07-19	CRITICAL	165.6	+327.1%	Spend above the day-of-week baseline by USD 2,229.69 (327.1%).
2026-07-27	CRITICAL	47.2	+87.9%	Spend above the day-of-week baseline by USD 635.37 (87.9%).
2026-07-26	LOW	-3.7	-7.1%	Spend below the day-of-week baseline by USD 50.33 (7.1%).

Top recommendations

• Sustained baseline eligible for commitment pricing — $4,919.93/mo, effort MEDIUM
• Idle weekend spend in EC2 — $4,731.92/mo, effort LOW
• Investigate spend above the modelled baseline — $2,772.64/mo, effort MEDIUM
• Idle weekend spend in RDS — $1,628.87/mo, effort LOW
• Spend concentrated in EC2 — $1,224.27/mo, effort MEDIUM

Generated by cloudsealed-jit — rolling-median baseline + MAD z-score, method in METHODOLOGY.md.