Refactor to make encoding more consistent

[AlexandreYang]

What does this PR do?

1. Use pymqi.ensure_bytes for better consistency with pymqi own internal use of pymqi.ensure_bytes.

Details:

#6913 (comment)

2. to_native_string is more appropriate than ensure_unicode.

Currently, in Python2 the code might actually break if there is a unicode in the channel name (unlikely though):

This code fails:

python2 -c 'tag = "abc: {}".format(u"12\xe1")'

Motivation

• Consistency
• Use native strings in majority of the code, and convert appropriately at the edge e.g. when interfacing with pymqi that require bytes.

[codecov]

Codecov Report

Merging #6995 into master will increase coverage by 2.73%.
The diff coverage is 100.00%.

Impacted Files	Coverage Δ
...ecks/ibm_mq/collectors/channel_metric_collector.py	88.60% <100.00%> (ø)
...checks/ibm_mq/collectors/queue_metric_collector.py	91.40% <100.00%> (ø)
solr/datadog_checks/solr/__init__.py
couch/tests/conftest.py
ibm_db2/datadog_checks/ibm_db2/utils.py
envoy/tests/test_envoy.py
consul/datadog_checks/consul/__init__.py
directory/datadog_checks/directory/directory.py
...ndry_api/datadog_checks/cloud_foundry_api/utils.py
kube_apiserver_metrics/tests/common.py
... and 781 more