elf: use relocations to find number of dynamic symbols if hash is empty

[philipc]

Fixes #111

[m4b]

Don’t we need to append the number of reloc syms found? E.g. can’t dynsyms be non-zero and we have relocs pointing at syms (in which case we leave out some imported symbols)

[philipc]

gnu_hash_len and hash_len already include imported symbols too, because imported symbols are always at the start of the table before the exported symbols. The problem is only when there are no exported symbols. If you want, we could do a max(relocs, gnu_hash_len), but it should not be necessary.

[m4b]

If you think max isn't necessary, then don't add it; but i just hope it's true :) i guess we don't really know about the probably of a binary like that in the wild, or if it even makes sense to parse it?

[philipc]

I may have been wrong about the imported symbols always being at the start... I couldn't find anything that says that. I've added another commit to always use the reloc symbols.

[m4b]

Better safe than sorry :)