slightly loosen ProgressMeter compat

[oscardssmith]

We want to avoid 1.10.3, but I don't think there is a strong reason to prevent future patch releases.

[github-actions]

Benchmark Results (Julia v1)

Time benchmarks

	master	1d053af...	master / 1d053af...
search/multithreading	17.6 ± 0.37 s	17.4 ± 0.23 s	1.01 ± 0.025
search/serial	32 ± 0.15 s	31.5 ± 0.22 s	1.02 ± 0.0086
utils/best_of_sample	1.69 ± 0.3 μs	1.8 ± 0.41 μs	0.939 ± 0.27
utils/check_constraints_x10	14.3 ± 3 μs	14.3 ± 2.9 μs	0.997 ± 0.29
utils/compute_complexity_x10/Float64	2.29 ± 0.14 μs	2.3 ± 0.14 μs	0.992 ± 0.086
utils/compute_complexity_x10/Int64	2.22 ± 0.15 μs	2.19 ± 0.11 μs	1.01 ± 0.085
utils/compute_complexity_x10/nothing	1.81 ± 0.16 μs	1.79 ± 0.13 μs	1.01 ± 0.12
utils/insert_random_op_x10	5.99 ± 1.9 μs	5.95 ± 1.9 μs	1.01 ± 0.46
utils/next_generation_x100	0.468 ± 0.023 ms	0.465 ± 0.022 ms	1.01 ± 0.068
utils/optimize_constants_x10	0.0348 ± 0.0084 s	0.0337 ± 0.0077 s	1.03 ± 0.34
utils/randomly_rotate_tree_x10	6.43 ± 0.68 μs	6.59 ± 0.75 μs	0.976 ± 0.15
time_to_load	2.32 ± 0.018 s	2.32 ± 0.01 s	0.998 ± 0.009

Memory benchmarks

	master	1d053af...	master / 1d053af...
search/multithreading	0.205 G allocs: 53.4 GB	0.206 G allocs: 53.3 GB	1
search/serial	0.207 G allocs: 55.2 GB	0.207 G allocs: 55.2 GB	1
utils/best_of_sample	0.037 k allocs: 3.22 kB	0.037 k allocs: 3.22 kB	1
utils/check_constraints_x10	0.037 k allocs: 2.33 kB	0.037 k allocs: 2.33 kB	1
utils/compute_complexity_x10/Float64	2 allocs: 0.969 kB	2 allocs: 0.969 kB	1
utils/compute_complexity_x10/Int64	2 allocs: 0.969 kB	2 allocs: 0.969 kB	1
utils/compute_complexity_x10/nothing	2 allocs: 0.969 kB	2 allocs: 0.969 kB	1
utils/insert_random_op_x10	0.047 k allocs: 2.84 kB	0.047 k allocs: 2.84 kB	1
utils/next_generation_x100	4.63 k allocs: 0.277 MB	4.63 k allocs: 0.277 MB	1
utils/optimize_constants_x10	25.5 k allocs: 26.7 MB	22.8 k allocs: 23.2 MB	1.15
utils/randomly_rotate_tree_x10	0.043 k allocs: 2.3 kB	0.043 k allocs: 2.3 kB	1
time_to_load	0.159 k allocs: 11.2 kB	0.159 k allocs: 11.2 kB	1

[MilesCranmer]

Unfortunately ProgressMeter.jl does not follow SemVer: https://github.com/timholy/ProgressMeter.jl/releases/tag/v1.10.1 (multiple features added on a patch release) so I think we should treat every patch release as breaking. Especially since it's such a key dependency in both SymbolicRegression.jl and PySR, I really don't think we can afford such a risk.

[oscardssmith]

@timholy would it be possible for ProgressMeter to more closely semver things? (specifically making minor releases for new features).

[Moelf]

@JamesWrigley

[timholy]

@MarcMush (new features should use a minor version bump)

[JamesWrigley]

@JamesWrigley

(just to clarify I'm not a maintainer of ProgressMeter, just a drive-by contributor 😛 )

[MarcMush]

@timholy yes I will be more careful