Finalize some experimental functions before 1.0

[szhorvat]

Experimental functions should be reviewed, and decided if they still stay experimental after 1.0. The procedure is simply to remove \experimental from the docs.

Suggestions by @szhorvat:

Drop experimental

• igraph_tree_from_parent_vector
• igraph_mean_degree — assuming we don't want to return the result directly
• igraph_count_loops — assuming we don't want to return the result directly
• igraph_get_shortest_path_astar
• igraph_hypercube — recent addition, but very straightforward
• igraph_is_clique
• igraph_is_independent_vertex_set
• igraph_hexagonal_lattice
• igraph_triangular_lattice
• igraph_joint_degree_matrix
• igraph_joint_degree_distribution
• igraph_joint_type_distribution
• igraph_enter_safelocale — warning: likely very little use in the wild
• igraph_exit_safelocale — warning: likely very little use in the wild
• igraph_distances_floyd_warshall
• igraph_distances_cutoff
• igraph_distances_dijkstra_cutoff
• igraph_graph_power
• igraph_join
• igraph_transitive_closure — recent addition, but very straightforward

To be discussed

• igraph_degree_correlation_vector
• igraph_is_complete — do we want to add a directed parameter?
• igraph_ecc — @vtraag, please review the offset parameter
• igraph_layout_merge_dla — wasn't this an old function, why experimental?
• igraph_layout_umap, igraph_layout_umap_3d, igraph_layout_umap_compute_weights — to be decided by @iosonofabio

Not ready to be dropped

• igraph_chung_lu_game — I am not entirely happy with the variant names
• igraph_minimum_cycle_basis — API not yet suitable for weighted generalization, needs igraph_real_t cutoff
• igraph_fundamental_cycles — API not yet suitable for weighted generalization, needs igraph_real_t cutoff; do we want weights at all?
• igraph_community_voronoi — give it more time
• all bitset functions — give it more time, recall that while the API is similar to vectors, the behaviour is different, actively being tested by making more and more use of it in the library
• igraph_reachability, igraph_count_reachable — give it more time