fix(firehose): seed firehose-apps.json with empty data; atomic write …

[castrojo]

…for images.json

• Replace 1.5MB populated firehose-apps.json with a minimal empty seed {"apps":[],"metadata":{"schemaVersion":"1.0.0","generatedAt":null}} so a clean git clone builds successfully without running fetch-firehose.js Closes docs(tips): my recommended extensions #48
• Write images.json atomically via a .tmp file + fs.renameSync to prevent build-time reads from seeing a partially-written file on concurrent runs Closes feat: move to Docusaurus! #49

Assisted-by: Claude Sonnet 4.6 via GitHub Copilot

[gemini-code-assist]

Code Review

This pull request modifies scripts/fetch-github-images.js to implement atomic file writes by using a temporary file before renaming it to the final destination. A review comment identified a potential race condition where concurrent executions of the script would collide on the same static temporary filename and suggested incorporating the process ID into the filename to ensure uniqueness.

[gemini-code-assist]

Using a static filename for the temporary file (OUTPUT_FILE + ".tmp") introduces a race condition if multiple instances of this script run concurrently. While the atomic rename protects readers, concurrent writers will collide on the same temporary file, potentially leading to data corruption or process crashes (e.g., when one instance renames the file while another is still writing to it). To safely handle concurrent runs as intended, use a unique temporary filename by incorporating the process ID.

Suggested change

	const TMP_FILE = OUTPUT_FILE + ".tmp";
	fs.writeFileSync(TMP_FILE, JSON.stringify(output, null, 2), "utf-8");
	fs.renameSync(TMP_FILE, OUTPUT_FILE);
	const TMP_FILE = OUTPUT_FILE + "." + process.pid + ".tmp";
	fs.writeFileSync(TMP_FILE, JSON.stringify(output, null, 2), "utf-8");
	fs.renameSync(TMP_FILE, OUTPUT_FILE);