Fix corpora naming

[capuanob]

Pull request

This Pull-Request includes the necessary changes to integrate fuzzing into pdfminer.six for OSS-Fuzz continuous fuzzing, as discussed in Issue 918.

In short, this PR adds atheris (the fuzzing framework) as a development dependency, and a new fuzzing directory containing a corpus, some initial harnesses, the necessary CI file to integrate the project into OSSFuzz, and a build script to be used by ClusterFuzz to prepare for nightly fuzzing.

In addition to the above, two simple bug-fixes are resolved to address crashes that were occurring too early into fuzzing, preventing progress.

How Has This Been Tested?

The fuzzing harnesses are tests in and of themselves, so they were tested via coverage analysis and allowing them to run.

NOTE: The CIFuzz.yml job will fail until Google merges the necessary pdfminer Dockerfile into their OSS-Fuzz repository. This can only be done after this PR is merged.

Checklist

• I have read CONTRIBUTING.md.
• I have added a concise human-readable description of the change to CHANGELOG.md.
• I have tested that this fix is effective or that this feature works.
• I have added docstrings to newly created methods and classes.
• I have updated the README.md and the readthedocs documentation. Or verified that this is not necessary.