Add .md to README and clarify what example does

[mjcross]

1. For pico_w/wifi/mqtt, rename README to README.md to improve how the content renders in an IDE or when browsing the online repository.
2. Clarify that the example doesn't re-publish the temperature reading if it hasn't changed; so that users can understand why the updates can appear sporadic.
3. Document how the led status can be checked by subscribing to /led/status

[peterharperuk]

Looks good. Just the one typo pointed out by @lurch

[lurch]

FYI #498 is where this example was first added.

[mjcross]

Thanks @peterharperuk and well spotted @lurch - fixed in my a23bca5. One of these days I'll manage to get a PR through without a typo 8-)

[mjcross]

If the dns query returns immediately then start_client() is called directly from main(); so shouldn't the call to mqtt_client_new() be protected? The lwIP implementation has LWIP_ASSERT_CORE_LOCKED

[mjcross]

With the greatest respect, although this example is very thorough it is also quite hard to follow.
Might I suggest that In the spirit of the other examples it would be clearer to use global variables and move some of the work outside the callbacks?

[peterharperuk]

If the dns query returns immediately then start_client() is called directly from main(); so shouldn't the call to mqtt_client_new() be protected? The lwIP implementation has LWIP_ASSERT_CORE_LOCKED

Yes

[peterharperuk]

If you have suggestions for other improvements and fixes. I would suggest raising them as separate pull requests.

[mjcross]

If the dns query returns immediately then start_client() is called directly from main(); so shouldn't the call to mqtt_client_new() be protected? The lwIP implementation has LWIP_ASSERT_CORE_LOCKED

Yes

The code as stands also calls mqtt_client_is_connected(), which the lwIP code also has the CORE_LOCKED assertion. Not sure whether that's safe or not?