More Verbose Token Information

### Describe the feature or problem you'd like to solve

Current token usage (/usage) only displays input, output, and cache tokens, along with the .jsonl files only displaying total token usage while prompts run.. Meanwhile claude cli is able to show their token usage as input, output, cache_read, cache_write. These numbers are also available in their session logs.

### Proposed solution

## Current
/usage
● Total usage est: 3 Premium requests
API time spent: 17s
Total session time: 1m 22.092s
Total code changes: +0 -0
Breakdown by AI model:
gpt-5.1 18.6k in, 59 out, 0 cached (Est. 1 Premium request)
claude-sonnet-4.5 102.0k in, 475 out, 50.2k cached (Est. 2 Premium requests)
## Proposed
● Total usage est: 3 Premium requests
API time spent: 17s
Total session time: 1m 22.092s
Total code changes: +0 -0
Breakdown by AI model:
gpt-5.1 18.6k in, 59 out, 0 read, 0 write (Est. 1 Premium request)
claude-sonnet-4.5 102.0k in, 475 out, 35.7k read, 14.5k write(Est. 2 Premium requests)

## Proposed additional logging
"usage":{"input_tokens":10,"cache_creation_input_tokens":5381,"cache_read_input_tokens":13513,"cache_creation":{"ephemeral_5m_input_tokens":5381,"ephemeral_1h_input_tokens":0},"output_tokens":3,"service_tier":"standard"}

### Example prompts or workflows

_No response_

### Additional context

_No response_