llms.txt links Explorer's landing stub instead of its section aggregate

## Current state

`layouts/index.llmstxt.txt:33` links InfluxDB 3 Explorer's per-page landing twin:

```
- [InfluxDB 3 Explorer](influxdb3/explorer/index.md): Web-based data exploration tool
```

Every sibling product in the InfluxDB 3 block (lines 28–31: Core, Enterprise, Cloud Dedicated, Cloud Serverless) links its `index.section.md` aggregate, and the template's preamble states the file points to aggregated `.section.md` files.

Verified against production (2026-06-11):

- `https://docs.influxdata.com/influxdb3/explorer/index.md` is the landing page only (`estimated_tokens: 303`).
- `https://docs.influxdata.com/influxdb3/explorer/index.section.md` exists (HTTP 200, ~14,400 estimated tokens, 10 child pages).

An agent that consumes products through `llms.txt` section entries receives ~300 tokens for Explorer and full section content for every other InfluxDB 3 product.

## Recommended changes

1. Change the entry to `influxdb3/explorer/index.section.md`.
2. Optional hardening: `yarn check:md-coherence` validates the "Full corpora" block against `getCorpusPaths()` but doesn't validate the hand-written section-block links. Add a check that each section entry links an existing file so the next hand-edited line can't drift.

## How this was found

Surfaced independently by two automated AI-visibility review runs against `/influxdb3/explorer/` (2026-06-11); both verified the production URLs above.

Open layouts/index.llmstxt.txt at line 33 and compare the Explorer entry with the sibling InfluxDB 3 section links and the preamble. Change the link to influxdb3/explorer/index.section.md, then run yarn check:md-coherence; completion means Explorer points to the existing aggregate, with the optional check covering hand-written section links.

Written by the indexing model from the issue text.