HDDS-4576. TestDefaultCertificateClient misuses chars param of random()

[adoroszlai]

What changes were proposed in this pull request?

TestDefaultCertificateClient passes "UTF-8" to RandomStringUtils.random(int, String) with the intention to use UTF-8 encoding. It was added in HDDS-1087 to fix CI failures.

However, the parameter is not for character encoding, rather:

     * @param chars  the String containing the set of characters to use,
     *  may be null, but must not be empty

So this results in values like: "--8F8T8U8T...".

This PR proposes another fix:

1. Allow random() to generate strings using all characters.
2. Fix test failure by specifying UTF-8 encoding when calling String.getBytes().

https://issues.apache.org/jira/browse/HDDS-4576

How was this patch tested?

Reproduced HDDS-1087 by removing chars argument:
https://github.com/adoroszlai/hadoop-ozone/runs/1530939654#step:3:3638

Verified the new fix:
https://github.com/adoroszlai/hadoop-ozone/runs/1530664863#step:3:3779

[xiaoyuyao]

LGTM, +1.

[adoroszlai]

Thanks @xiaoyuyao for reviewing and committing it.