More consistent logging

[fabianegli]

Description of feature

There seems to be undesirable variation in how the nf_core/main.py script handles logging before the various sys.exit(1) codes. i.e. sometimes no logging, sometimes logging to log.error and sometimes to log.critical.

This is a list of except -> ... -> sys.exit(1) code snippets (sys.exit(1) is not shown for brevity):

240
    except LookupError as e:
        log.error(e)

320
    except UserWarning as e:
        log.error(e)

331
    except AssertionError as e:
        log.critical(e)

334
    except UserWarning as e:
        log.error(e)

361
    except LookupError as e:
        log.critical(e)

389
    except UserWarning as e:
        log.critical(e)

414
    except UserWarning as e:
        log.critical(e)

445
    except UserWarning as e:
        log.error(e)

494
    except UserWarning as e:
        log.error(e)

518
    except UserWarning as e:
        log.critical(e)

560
    except UserWarning as e:
        log.critical(e)

583
    except UserWarning as e:
        log.critical(e)

613
    except nf_core.modules.lint.ModuleLintException as e:
        log.error(e)

616
    except UserWarning as e:
        log.critical(e)

648
    except UserWarning as e:
        log.error(e)

668
    except nf_core.modules.module_utils.ModuleException as e:
        log.error(e)

671
    except UserWarning as e:
        log.critical(e)

727
    except UserWarning as e:
        log.critical(e)

745
    except AssertionError as e:
        log.error(e)

749
    except AssertionError as e:
        [no logging]

786
    except (UserWarning, AssertionError) as e:
        log.error(e)

813
    except AssertionError as e:
        [no logging]

849
    except AssertionError as e:
        log.error(e)

893
    except UserWarning as e:
        log.error(e)

934
    except (nf_core.sync.SyncException, nf_core.sync.PullRequestException) as e:
        log.error(e)

There are also incidents of sys.exit(1) after an if statement. Also with seemingly inconsistent logging.

[ewels]

log.error / log.critical - yeah I don't have a strong opinion on which should be used, if you'd like to homogenise them then that's fine.

No logging - this is very likely because there is logging coming from the module code. So I doubt any addition is needed in the cli handling code.