code improvement for DrugMechDB

[eKathleenCarter]

          [drop_duplicates](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.drop_duplicates.html)

is more efficient for lines 197 and 220

I would suggest replacing lines 198 through 216 with the following:

df.rename(columns={"dmdb_ids": "drugmechdb_path_id",
"qualified_predicates": QUALIFIED_PREDICATE,
"object_direction_qualifiers": OBJECT_DIRECTION_QUALIFIER,
"object_aspect_qualifiers": OBJECT_ASPECT_QUALIFIER},
inplace=True)
df[KNOWLEDGE_LEVEL] = KNOWLEDGE_ASSERTION
df[AGENT_TYPE] = MANUAL_AGENT

df['edge_props'] = df.apply(lambda x: x[QUALIFIED_PREDICATE,
OBJECT_DIRECTION_QUALIFIER,
OBJECT_ASPECT_QUALIFIER,
KNOWLEDGE_LEVEL,
AGENT_TYPE].dropna().to_dict(), axis=1)

for i, row in df.iterrows():

output_edge = kgxedge(
    subject_id=row["source_ids"],
    object_id=row["target_ids"],
    predicate=row["predicates"],
    edgeprops=row['edge_props'],
    primary_knowledge_source=self.provenance_id
)
self.output_file_writer.write_kgx_edge(output_edge)

Because iterrows is EXTREMELY slow and inefficient

Originally posted by @eKathleenCarter in #221 (comment)