docs: update pipedream docs

[daveomri]

Updates the Pipedream integration doc to match the connector changes from the retest, plus some housekeeping.

What changed

• Copy: removed the deleted Run Task Synchronously action; updated Run Actor params (schema-driven input, Memory dropdown, Build accepts tag or number, Wait for Finish switch); Get Dataset Items now notes ID or name.
• Screenshots: refreshed the Run Actor screenshot and converted all images to WebP (dropped the PNGs).
• Logo + hub cards: added the official Pipedream SVG logo and a Pipedream card to both integration hubs (it was previously missing from the hubs; the old PNG logo was unreferenced).
• AI usage: added a note pointing to the Apify MCP server for AI agents, and a caution that Pipedream's AI builder is token-heavy for this schema-driven connector.

Note

** Merge this alongside the connector release**, not before.

[marcel-rbro]

Blocked on the connector release.

[marcel-rbro]

Can you confirm the engine list? On master the crawlerType prop has four options and this covers three, missing playwright:chrome — the suggestion assumes it's still there and uses the app's own labels. The adaptive option's label is a full sentence in the app, so there's no short name to bold.

Suggested change

- **Scrape single URL** - Runs [Website Content Crawler](https://apify.com/apify/website-content-crawler) on a specified URL and returns its content as HTML. Use this for extracting content from a single page, e.g. in LLM workflows. **Crawler Type** selects the engine: **Firefox (Headless Browser)** renders JavaScript (the default), **Cheerio (Raw HTTP)** sends raw HTTP requests and renders none, and **Adaptive** switches per page.

- **Scrape single URL** - Runs [Website Content Crawler](https://apify.com/apify/website-content-crawler) on a specified URL and returns its content as HTML. Use this for extracting content from a single page, e.g. in LLM workflows. **Crawler Type** selects the engine: **Headless browser (stealthy Firefox+Playwright)** is the default, **Headless browser (Chrome+Playwright)** also renders JavaScript, **Raw HTTP client (Cheerio)** sends raw HTTP requests and renders none, and the adaptive option switches between raw HTTP and Chrome per page.

[marcel-rbro]

Where does the 256 KB threshold come from, Pipedream's step output or Apify's key-value store? Readers need to know whose limit it is to know what to change, and naming the reference object's fields here grounds the recordUrl and recordKey used just below.

Suggested change

	When **Run Actor** waits for a run, it returns the record named by **Output Record Key** (`OUTPUT` by default). If the record is over 256 KB, the step returns a reference instead of inline data.
	When **Run Actor** waits for a run, it returns the record named by **Output Record Key** (`OUTPUT` by default). If the record is over 256 KB, the step returns a reference object with `recordUrl`, `keyValueStoreId`, and `recordKey` instead of inline data.

[marcel-rbro]

Would it be better to name the terminal states here? They're only listed under Triggers further down, so at this point the reader can't tell what they'd be opting out of.

Suggested change

	1. Configure the trigger by selecting the Actor or task to monitor. Leave **Trigger on run states** empty to fire on every terminal state, or pick the states you care about.
	1. Configure the trigger by selecting the Actor or task to monitor. Leave **Trigger on run states** empty to fire on every terminal state (**Succeeded**, **Failed**, **Timed out**, and **Aborted**), or pick the states you care about.