Automatically suggest similar parameters

[MilesCranmer]

PySRRegressor has a lot of parameters. To help with this, this PR implements an "autosuggestion" feature that, upon receiving an invalid keyword, will print an error message with possibilities for what the user had meant to write out.

In [2]: model = PySRRegressor(iterations=10)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[2], line 1
----> 1 model = PySRRegressor(iterations=10)

File ~/PermaDocuments/SymbolicRegressionMonorepo/PySR/pysr/sr.py:916, in PySRRegressor.__init__(self, model_selection, binary_operators, unary_operators, niterations, populations, population_size, max_evals, maxsize, maxdepth, warmup_maxsize_by, timeout_in_seconds, constraints, nested_constraints, elementwise_loss, loss_function, complexity_of_operators, complexity_of_constants, complexity_of_variables, parsimony, dimensional_constraint_penalty, dimensionless_constants_only, use_frequency, use_frequency_in_tournament, adaptive_parsimony_scaling, alpha, annealing, early_stop_condition, ncycles_per_iteration, fraction_replaced, fraction_replaced_hof, weight_add_node, weight_insert_node, weight_delete_node, weight_do_nothing, weight_mutate_constant, weight_mutate_operator, weight_swap_operands, weight_randomize, weight_simplify, weight_optimize, crossover_probability, skip_mutation_failures, migration, hof_migration, topn, should_simplify, should_optimize_constants, optimizer_algorithm, optimizer_nrestarts, optimize_probability, optimizer_iterations, perturbation_factor, tournament_selection_n, tournament_selection_p, procs, multithreading, cluster_manager, heap_size_hint_in_bytes, batching, batch_size, fast_cycle, turbo, bumper, precision, enable_autodiff, random_state, deterministic, warm_start, verbosity, update_verbosity, print_precision, progress, equation_file, temp_equation_file, tempdir, delete_tempfiles, update, output_jax_format, output_torch_format, extra_sympy_mappings, extra_torch_mappings, extra_jax_mappings, denoise, select_k_features, **kwargs)
    914 if len(suggested_keywords) > 0:
    915     err_msg += f" Did you mean {' or '.join(suggested_keywords)}?"
--> 916 raise TypeError(err_msg)

TypeError: iterations is not a valid keyword argument for PySRRegressor. Did you mean niterations or migration or optimizer_iterations?

Note the new addition to the error message:

Did you mean niterations or migration or optimizer_iterations?

It does this using difflib and suggests up to 3 options (but only if they are past a threshold similarity).

To be merged after #609 as it is rebased on top of that.