Harden text parsers against malformed fuzz inputs

[Copilot]

Several whitespace-delimited loaders were still crashing on malformed freeform input instead of failing soft. In practice, blank lines, comment lines, and underspecified rows could trigger raw indexing/unpacking errors across multiple parser entry points.

• Parser hardening

  • Skip blank and comment lines consistently in text-based parsers.
  • Treat underspecified rows as non-records instead of raising IndexError/ValueError.
  • Add default-weight handling where short-but-valid rows are accepted.

• Formats covered

  • multiedgelist
  • simple edgelist
  • edgelist_with_edge_types
  • spin_edgelist
  • multiplex_edges
  • multiplex folder readers for edge/activity/layer files

• Behavioral correction

  • parse_edgelist_multi_types() now accepts 3-column rows (node1 node2 weight) without assuming a fourth edge_type field.

• Regression coverage

  • Added focused tests for malformed-but-common fuzz inputs:
    • leading blank lines
    • comment-prefixed lines
    • short rows with omitted optional fields

Example of the now-supported soft-fail/default behavior:

# accepted: optional weight only
p q 3.0

# skipped safely
# comment