fix(core): small changes to diagnostics

[allardy]

1. Sometime the list of redis clients is not displayed, maybe there's a problem with the parsing. This change will still return the raw list so we can see what is the issue
2. Added a try/catch block around the database tables logic just in case
3. List the number of messaging nodes listening on redis (useful to debug rogue processes)
4. Checks the connection between bp and the other processes (studio/messaging/nlu).

If no endpoint is specified, it will try to connect on different hosts:

• localhost
• 127.0.0.1
• the server's hostname
• external address configured

This could help determine connection issues in various settings (eg: restrictions on a specific hostname, etc).

@franklevasseur @samuelmasse @laurentlp Can you check if the logic I use there is correct, and do you think something else can be added to check the overall health of everything?

[samuelmasse]

List the number of messaging nodes listening on redis (useful to debug rogue processes)

Doesn't messaging already do this with Registered foreign node x?

[allardy]

@samuelmasse The diagnostic report takes data from various places to give a clear picture of the client's situation. It's a summary of its installation. And since we don't have messaging logs steamed to the admin, in the log file or in the db, it's necessary to consolidate the information there because we don't often have access to the infra guy who can check the stdout

[samuelmasse]

LGTM

[samuelmasse]

Why does this need to be added here?

[allardy]

To be effective the diag must avoid using any of botpress' internals and reflect the true state of the app. The nlu endpoint can come from the module config, or from 2 env variables, so there's no way to be 100% sure of what the effective value is (and I cant use the ghost to read the config in the db)

[samuelmasse]

@allardy can we merge this?