docs: Add token consumption reference table to README

[Copilot]

Users had no baseline to estimate API costs before running a simulation.

Changes

• Added a Token Consumption Reference section in the Quick Start area with a measured data table (100 agents, activation probability 1, 1 time step, using QWEN_TURBO: 335,600 input tokens / 16,750 output tokens)
• Added a callout note reminding users that token usage scales with agent count, activation probability, and time steps
• Added a Qwen pricing table for estimated costs at 1 time step, activation probability 0.1 (Qwen pricing as of 2024-12-14), covering qwen-plus and qwen-max models across 100, 1,000, and 10,000 agents

[echo-yiyiyi]

@copilot Could you fix the pre-commit?

[Copilot]

@copilot Could you fix the pre-commit?

Fixed in a4ed215 — mdformat was reformatting the table with aligned column widths. The pre-commit hooks all pass now.

[echo-yiyiyi]

@copilot Could you also add these data, 运行1个时间步，激活概率0.1估算价格(12/14/2024 Qwen模型调用定价)：

Agent数量	100	1k	1w
qwen-plus	0.026848元	0.26848元	2.6848元
qwen-max
0.717元	7.717元	77.17元

[Copilot]

@copilot Could you also add these data, 运行1个时间步，激活概率0.1估算价格(12/14/2024 Qwen模型调用定价)：
Agent数量 | 100 | 1k | 1w
-- | -- | -- | --...

Added in 3f3341d. The pricing table for qwen-plus and qwen-max (1 time step, activation probability 0.1, Qwen pricing as of 2024-12-14) now appears below the token consumption reference:

Model	100 Agents	1,000 Agents	10,000 Agents
qwen-plus	¥0.026848	¥0.26848	¥2.6848
qwen-max	¥0.717	¥7.717	¥77.17

All pre-commit hooks pass.

[echo-yiyiyi]

lgtm. Thanks @copilot