Fix flycheck-error-level-interesting-p

[KevOrr]

According to its docstring, flycheck-error-level-interesting-p should return true if there are no errors as or more severe than flycheck-navigation-minimum-level. Currently, however, it only checks if there are any errors exactly as severe as flycheck-navigation-minimum-level (via flycheck-has-current-errors-p). This patch:

• improves the documentation of
  • flycheck-has-current-errors-p
  • flycheck-has-errors-p
• introduces two new functions
  • flycheck-has-current-errors-atleast-p
  • flycheck-has-errors-atleast-p
• Changes flycheck-error-level-interesting-p to use flycheck-has-current-errors-atleast-p, instead of its "exact" sibling, bringing its actual behavior in line with its documented behavior

I also considered changing flycheck-has-[current-]errors-p to use >= instead of eq, but since it has a public name (I'm assuming -- is used to signal that a symbol is private), I didn't want to introduce a backwards-breaking change. If instead it should be the case that flycheck-has-[current-]errors-p check for errors at least as severe as level, then I can edit this PR to modify them directly.

[CLAassistant]

All committers have signed the CLA.

[KevOrr]

Oops, I should have run the tests locally before submitting this PR. It seems as though some of the tests rely on the old behavior of flycheck-error-level-interesting-p. For example flycheck-first-error/over-warnings/goes-to-first-error assumes that by setting flycheck-navigation-minimum-level to 'warning, then (flycheck-first-error) will move onto the checkdoc error of level 'info on line 12 of errors-and-warnings.el. Should the checkdoc error indeed be of level 'info or is this a local configuration thing, and it should be 'warning or higher?

[bbatsov]

Seems like a duplicate of #1994. We should agree which approach to solving the issue is better.