Bug: clamp being used incorrectly after hours running

[johnnyasantoss]

Hello,

I couldn't create an issue so I'll be using this discussion thread to report a bug.

Yesterday I was testing and reviewing #33, after sending the review I left my bitaxe 602 mining the whole night on that branch to gather more data to plot (it was pretty stable after modifications).
I woke up and found this log:

thread 'tokio-runtime-worker' (31152351) panicked at /Users/johnny/.rustup/toolchains/nightly-aarch64-apple-darwin/lib/rustlib/src/rust/library/core/src/cmp.rs:1094:9:
assertion failed: min <= max
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Unfortunately I was running without backtrace (shame), but after a quick repo wide search I only found one place were clamp is used with dynamic inputs:

mujina/mujina-miner/src/scheduler.rs

Lines 268 to 271 in 491acac

	let measurement_target = MEASUREMENT_SHARE_RATE.to_target(hashrate);
	let flood_cap_target = FLOOD_CAP_RATE.to_target(hashrate);
	source_target.clamp(measurement_target, flood_cap_target)

The miner continued mining but I'm not sure that it should. I'm not that familiar with the codebase to tell just by looking at this piece of code to say which thread was running this. Maybe an essential actor?

[rkuester]

Created #49 from this discussion.