Implement a debug mode to simplify quick bug analysis

Overview

We need a debug mode available because errors are caught in a few places in the framework. For example, the CLI catches all the errors or Report.from_validate.

We might use an env var or a flag

[fjuniorr]

@roll just to add additional motivation for this I’ve had a hard time tracking down a wrong delimiter (,) specified in the csv dialect of a resource when the actual csv file was using ;.

Here is an example of the message:

from frictionless import Package, validate_resource
from pprint import pprint

dp = Package('https://raw.githubusercontent.com/dados-mg/datapackage-reprex/bad-dialect-delimiter/datapackage.json')

report = validate_resource(dp.get_resource('pib-per-capita'))

pprint(report)

{'errors': [{'code': 'task-error',
             'description': 'General task-level error.',
             'message': "The task has an error: 'fieldPosition'",
             'name': 'Task Error',
             'note': "'fieldPosition'",
             'tags': []}],
 'stats': {'errors': 1, 'tasks': 0},
 'tasks': [],
 'time': 2.439,
 'valid': False,
 'version': '4.16.6'}

I think that this more obscure message has something to do with the fact that the resource pib-per-capita has a foreignKeys constraint, because the error in the other resource (ie. validate_resource(dp.get_resource('estados'))) is more clear.

Ps. I’ve found https://github.com/frictionlessdata/frictionless-py/issues/903 in the process of investigating this problem and initially though they could be related. 😅

[roll]

DONE in #1119