Fix #48, string table identification.

[jphickey]

Describe the contribution

Support ELF files that have all strings, including ELF section names, in one single ".strtab" section in the ELF file.

This gives preferential treatment to a section named ".strtab" as this should always be the section which contains the names for the ".symtab" section.

This should also be true of ELF files that contain other strtab sections, such as a ".dynstr" section.

Fixes #48

Testing performed
Confirmed that normal table files built using GCC (e.g. those in the sample framework configuration) are identical, before and after this change is applied.

Confirmed that this is now able to also correctly parse an ELF file from clang which has all strings in a single .strtab section.

Expected behavior changes
Now correctly parses ELF files that do not have a separate .shstrtab and .strtab section.

System(s) tested on
Ubuntu 20.04

Contributor Info - All information REQUIRED for consideration of pull request
Joseph Hickey, Vantage Systems, Inc.

[jphickey]

Note, this also adds a warning/error exit condition if no usable string table is identified.

In the previous code, it would proceed using offset 0 (the init value of StringTableDataOffset) but this was wrong, and as a result the string data was all garbage data as it read parts of the ELF header interpreted as strings. The program should exit with an error if this offset hasn't been set to a nonzero value by this point.

Additionally, if the tool failed to identify the CFE_TBL_FileDef symbol, it would print an error but still exit with zero status (success) even though no output was written. As a result the build continued as if nothing was wrong, but the table file is missing.