feat!: automatically detect format in read_graph() and write_graph()

[maelle]

Fix #511

Co-Authored-By: Claude Sonnet 4.6

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 9c72dda is merged into main:

• ✔️as_adjacency_matrix: 758ms -> 756ms [-0.73%, +0.16%]
• ✔️as_biadjacency_matrix: 735ms -> 734ms [-0.71%, +0.44%]
• ✔️as_data_frame_both: 1.47ms -> 1.48ms [-1.06%, +2.24%]
• ✔️as_long_data_frame: 4.08ms -> 4.08ms [-2.53%, +2.78%]
• ✔️es_attr_filter: 2.64ms -> 2.65ms [-0.48%, +1.16%]
• ✔️graph_from_adjacency_matrix: 117ms -> 116ms [-1.91%, +1.27%]
• ✔️graph_from_data_frame: 3.44ms -> 3.43ms [-3.08%, +1.96%]
• ✔️vs_attr_filter: 1.5ms -> 1.54ms [-1.61%, +6.62%]
• ✔️vs_by_name: 963µs -> 964µs [-1.71%, +1.76%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 1028a38 is merged into main:

• ✔️as_adjacency_matrix: 780ms -> 775ms [-3.12%, +1.97%]
• ✔️as_biadjacency_matrix: 762ms -> 760ms [-2.32%, +1.85%]
• ✔️as_data_frame_both: 1.46ms -> 1.47ms [-1.02%, +2%]
• ✔️as_long_data_frame: 3.85ms -> 3.85ms [-0.55%, +0.46%]
• ✔️es_attr_filter: 2.84ms -> 2.84ms [-1.68%, +2.02%]
• ✔️graph_from_adjacency_matrix: 123ms -> 122ms [-2.38%, +1.24%]
• ✔️graph_from_data_frame: 3.29ms -> 3.3ms [-0.41%, +0.84%]
• ✔️vs_attr_filter: 1.65ms -> 1.66ms [-2.79%, +3.14%]
• ✔️vs_by_name: 1.15ms -> 1.09ms [-15.45%, +4.45%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 45baa3e is merged into main:

• ✔️as_adjacency_matrix: 751ms -> 745ms [-3.01%, +1.34%]
• ✔️as_biadjacency_matrix: 737ms -> 736ms [-1.15%, +0.87%]
• ✔️as_data_frame_both: 1.49ms -> 1.51ms [-0.59%, +3.38%]
• ✔️as_long_data_frame: 3.85ms -> 3.83ms [-1.98%, +0.61%]
• ✔️es_attr_filter: 2.73ms -> 2.73ms [-1.36%, +1.22%]
• 🚀graph_from_adjacency_matrix: 144ms -> 142ms [-3.77%, -0.18%]
• ✔️graph_from_data_frame: 3.37ms -> 3.34ms [-2.52%, +0.85%]
• ✔️vs_attr_filter: 1.53ms -> 1.53ms [-2.41%, +2.54%]
• ✔️vs_by_name: 962µs -> 970µs [-1.11%, +2.93%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[maelle]

Question for the hackathon: how should I handle the tests' dependency on igraphdata?