feat: import performance

mainly the change of the motifs md5 comparison (looking for page over-spanning motifs) decreases the import timing. instead of doing a tree traversal, we now compare sequences of md5 sums. this is much faster. other changes here are comments based on profiling. also index configuration has been tested, but did not work or provide any better performance, most likely since we are comparing nodes that are already in the built-in structural index and no further optimization is possible. furthermore one depended has been updated since it also got a performance boost based on requirements regarding this application.

in addition this MR adds support for the 'AHIQAR_TEXTAPI_URL' env var to determine the endpoint.