fix(web): clarify token usage labels in message details

[techotaku39]

Problem

The message-details popover labels input_tokens + output_tokens as "billable tokens":

Usage: 125 billable tokens (100 in / 25 out)

That wording makes a billing claim the UI cannot reliably support across providers. The available message metadata only describes token counts; provider-specific pricing can treat cache I/O and other token categories differently. The existing Usage / Total label also changes between single-turn and aggregated response groups even though both rows present the same token metric.

Solution

Use one neutral, explicit token label for both single-turn and aggregated message details:

Tokens: 125 total (100 in / 25 out)

The total remains exactly input_tokens + output_tokens; only the presentation changes. The existing input/output breakdown, number formatting, model/tier metadata, and multi-turn count remain unchanged.

Touch-device access needs no additional change here: current main already keeps the message info action reachable on touch devices via #1046.

Testing

• bun typecheck
• bun run test:cli
• bun run test:hub
• bun run test:web — 155 files, 1,259 tests passed
• bun run test:shared
• Focused message metadata/action tests — 19 tests passed
• git diff --check

Manual verification

Built the production web bundle and embedded hub assets, then served the result on a local HAPI hub. Verified that the emitted bundle contains the new Tokens: … total (… in / … out) format and no longer contains billable tokens.

[github-actions]

Findings
No findings.

Summary
Review mode: initial
No issues found in the changed lines. Residual risk: this was a static review only; I did not execute the PR tests in this automation pass.

Testing
Not run (automation)

HAPI Bot