set-random-seed added

[DaddyWesker]

So, I've figured out how to properly use thread_local for creating thread_local global variable in Rust. Usage of random-int and random-float is the same. Base seed is coming from StdRng::from_os_rng(). New set-random-seed is used to set a random seed. So when this operation is called, GLOBAL_RNG (which is thread local) will be reinitialized with a new seed.

I'm not sure what kind of tests should I provide for set-random-seed operation so I've created kinda simple one.

Example is quite simple:
!(random-int 0 5) # just a regular call without pre-defined seed as before
!(set-random-seed! 5) # setting a seed to a random generator
!(random-int 0 5) # now calling random-int after reinitializing generator with a seed.

Update. It was decided to refactor API and add some additional functions like new-random-generator etc. So API is know looks like this:

1. new-random-generator creates random generator instance for further usage. Takes seed as an argument;
2. random-int\random-float now takes not only range but also instance of random generator;
3. there is a metta-level (token) random generator with default behavior, e.g. StdRng::from_os_rng(); which called &rng and which can be used just like user-creeated random generator instance;
4. set-random-seed accepts number and random generator instance and changes its behavior using random seed;
5. reset-random-generator discards all changes to random generator instance and reinitialize it using StdRng::from_os_rng();.

Example of usage

!(bind! &newrng (new-random-generator 9)) # 9 is a seed for creating random generator instance
!(random-int &newrng 0 10) # using just created &newrng to generate a number
!(random-float &rng 0 10) # &rng is a global metta-level token
!(set-random-seed &newrng 5) # re-seeding random number generator instance
!(set-random-seed &rng 5) # we can also re-seed global &rng
!(reset-random-generator &rng) # resetting &rng to the default behavior

[vsbogd]

Please add a test which checks that set-random-seed leads to deterministic results. Also I would add a code proposed by @Necr0x0Der if you have one.