BUGFIX: fix globalHealth not returning to OK after a FAILED state

[joelsdc]

We have run into an issue where once a HC has FAILED, it will never go back to OK.

This is because on each healthcheck.Check() we scan for failed checks, if any check fails, we set the global health to false and return a 503. On subsequent checks, if all results are OK, there is not code to set the global health back to OK.

This PR ensures we start from an OK state on each check, if any checks fail, we set the global state to false. On subsequent requests, we set back to OK prior to doing the checks, so if any checks fail it will be set again to FALSE before replying, and if all checks are OK, the OK state is kept.

@adrianlop @jeevatkm please let me know your thoughts on this.

As this is a bugfix, we should tag a new version.

[adrianlop]

@joelsdc @jeevatkm please check #7

good catch btw 😄

[joelsdc]

PR updated with the changes @adrianlop proposed.

Please review and if it looks good we can merge and create a new release.

[jeevatkm]

@joelsdc Changes look good; we can merge them. I just noticed Travis CI build blocking the PR, and it's no longer free for open source. I think its better to use GitHub actions.