Add /installed_extensions endpoint to collect statistics about extension usage.

[lubennikovaav]

Add /installed_extensions endpoint to collect
statistics about extension usage.
It returns a list of installed extensions in the format:

{
  "extensions": [
    {
      "extname": "extension_name",
      "versions": ["1.0", "1.1"],
      "n_databases": 5,
    }
  ]
}

[github-actions]

5120 tests run: 4913 passed, 0 failed, 207 skipped (full report)

---

Flaky tests (3)

Postgres 17

• test_sharding_backpressure: release-x86-64

Postgres 16

• test_cannot_create_endpoint_on_non_uploaded_timeline: release-arm64

Postgres 15

• test_subscriber_restart: release-x86-64

Code coverage* (full report)

• functions: 31.4% (7510 of 23947 functions)
• lines: 49.5% (60295 of 121746 lines)

* collected from Rust tests only

---

_{The comment gets automatically updated with the latest test results
29f4e1a at 2024-10-09T11:32:00.283Z :recycle:}

[ololobus]

Left a bunch of comments. My main concern is that as we deal a lot with third party code and formatting, we'd better handle all errors gracefully and only use expect/unwrap for stuff that 'should never happen' -- mutexes, runtimes, etc.

[hlinnaka]

How is the new API endpoint going to be called? Who's going to call it and when?

Did you consider exporting this information as a prometheus metric, instead?

[lubennikovaav]

How is the new API endpoint going to be called? Who's going to call it and when?

In my idea the caller will be cplane, similar to GetStartupMetrics.

Did you consider exporting this information as a prometheus metric, instead?

Is it possible to add arbitrary hashmap as prometheus metric? I thought, it only supports prometheus metric types.