Streamline submission filtering and submission error handling

[tsaglam]

Currently, there are multiple ways how submissions are excluded. Some are removed during iterating while parsing (see JPlag.parseSubmissions()), while others are marked as erroneous and then later filtered in a later step. We should streamline this process.
As pointed out by @Alberth289346 in #230:

iter.remove() while parsing/checking submissions feels like the wrong approach. It's slow to remove in a long arraylist, and more importantly, it messes in the job that filterValidSubmissions() should do. I think this needs a decision where invalid submissions are removed. For paranoia checking, it might be useful to check further down the line that only valid submissions exist.

In my opinion, submissions should be marked as erroneous and maybe even given the information why they are erroneous. Then, they should be filtered later on. If we introduce a submission set (i.e. with #231), this could be done in there resulting in two methods: getValidSubmissions() and getInvalidSubmissions(). The latter is also useful for the new report generation, as the current JPlagResult does not expose which submissions were not used for comparison.