Fix nmap table result parsing

[botastic]

As mentioned in #9 the data from tables isn't always parsed correctly. As I was already checking the nmap DTD XML specifications, I saw that the relation between scripts, tables and elements is a bit more complicated.
While your approach - using a map for one table - was quiet comfortable it lacked completeness.

Here's an excerpt of the relevant section in the DTD:

<!ELEMENT script	(#PCDATA|table|elem)* >
<!ATTLIST script	
	id	CDATA	#REQUIRED
	output	CDATA	#REQUIRED
>

<!ELEMENT table	(table|elem)* >
<!ATTLIST table
    key CDATA #IMPLIED
>

<!ELEMENT elem	(#PCDATA)>
<!ATTLIST elem
    key CDATA #IMPLIED
>

The major problems are:
- nested tables
- elements without key
- (tables without key)

Because of the optional keys I decided to create new structs for Tables and Elements in my approach. I also considered adding an additional map to each one of the new structs, that would allow for a more comfortable access. But in the end I decide against that, as the missing keys would still be a problem.

Currently the script does not have a field for the PCDATA. This was not of relevance to me but can be easily added.

As in my other PR I adapted the tests and extended them a bit to better suit the new implementation. Additionally I also changed the TestToFile function, as in the other PR.

If you'd like to accept both PRs, I also have a branch that currently holds all the changes,

Best regards

PS: The first commits have been reverted, as I created the branch after submitting my first PR and wanted to create two PRs for an easier review ;)

[Ullaakut]

LGTM!