feat: add simulation config

[enigbe]

What this PR does

• Adds a SimulationConfig struct that Loads config values from an optional INI file and overwrites these values if corresponding command line arguments are passed
• The config file contains custom values for the simulation

  [simln.conf]
  data_dir = "."
  sim_file = "sim.json"
  total_time=
  print_batch_size = 500
  log_level = "info"
  expected_pmt_amt = 3800000
  capacity_multiplier = 2
  no_results = false
  log_interval = 60

• Makes configurable the log_interval after which results are logged

Related Issue(s)

• Resolves #157

Update

• This PR has changed to consider a standard way of handling configuration files as well as command line arguments. It currently introduces functionality to create and parse the Cli from an optionally-provided config file, to merge this Cli with that parsed from command line arguments, and to grant precedence to the options/values in the latter.

• Secondly, this PR also replaces the logging crate from simple_logger to flexi_logger. flexi_logger allows dynamic update of log levels after initialization and was required to deal with the issue described here.

• Finally, documentation was updated to reflect the introduction and use of a config file.

Note: There are some changes to unrelated files because of the requirement to wrap maximum width of code and comments at 120 column spaces. Thought to include them regardless following conversation from #164 on maintaining width.

[okjodom]

@enigbe here comes a late review!

I've read through the changes proposed and seem beyond the scope of #157

Add a SimulationConfig struct that passes all the configuration options to the simulator rather than individual params.

I was expecting we create this new struct in sim-lib and update Simulation::new() to take nodes, activity, and config as params. In sim-cli we can just add a type coercion which converts the cli params into SimulationConfig and use it after Cli.parse(). Example coercion below

impl Cli {
    fn to_simulation_config(&self) -> anyhow::Result<SimulationConfig> {
        Ok(SimulationConfig {
            total_time: self.total_time.unwrap_or(60),
            print_batch_size: self.print_batch_size,
            log_level: self.log_level,
            expected_pmt_amt: self.expected_pmt_amt,
            capacity_multiplier: self.capacity_multiplier,
            no_results: self.no_results,
        })
    }
}

async fn main() {
  let cli = Cli::parse();
  let opts = cli.to_simulation_config().unwrap();
  ...  
}

Any changes that require the .ini might make sense if there's a user scenario needing it. I'm not aware of any so deferring to @carlaKC

Make log_interval configurable while we're here (currently hard set to 60s)

cool beans on this part of the issue. change lgtm :)

[carlaKC]

While this is outside of the scope of the original issue, I think it's a really nice extension to simln / pretty standard way to handle configuration. I can see it being useful when using simln in larger deployments where you want to have a config file checked in rather than deal with cli flags, so my vote is to keep the proposed functionality 👍

I was expecting we create this new struct in sim-lib and update Simulation::new() to take nodes, activity, and config as params.

Same here - the original issue was opened because clippy gets unhappy with the number of parameters that Simulation:new takes, so we're going to want to add a struct to sim-lib.

[carlaKC]

Needs rebase!

[carlaKC]

Since this is pretty stale now and needs a rebase anyway, let's split it out into a change for just adding a simple config struct and another for a config file - as originally flagged in early review. Would also be nice to have some of the default mechanics in place that we have for the sim file (automatically detect a default value, for example).

[carlaKC]

Closing due to inactivity, please feel free to re-open if you've got capacity to pick this back up with the suggested split.