By only including selected tests we can skip checking for the
enabled/disabled status. This will also simplify the process
of disabling future tests for a disk if a failure is detected.
Addresses issue #184
The first read phase will skip a lot more to try to recover more data
from the whole source. Then the second read phase will fill in like
the previous configuration.
This is done to define the domain size and let us use --complete-only.
This also enables us to open ddrescueview immediately since that tool
requires a valid map file from the start. If you open an empty map
file ddrescueview doesn't auto-reload the file correctly.
Addresses #184
If the CPU reaches the failing temps during Prime95 then sysbench will be
run to emulate a less artificial workload. The both the overall and sysbench
max temps are recorded and shown in the results.
Added new option to track an alternate max temp value in wk.hw.sensors.
This was needed so show two different max temps recorded during CPU testing.
Sysbench was added to the Linux package list and is compiled for macOS.
Without manually compiling the package it brings in way too many dependencies
to support SQL DB testing (which we don't need).
Currently limited to only cloning whole disks, not select partitions.
We need to specify the --size due to a bug under macOS; it fails to
detect the size of the device/partition and reports 8192 PiB instead.
Before starting a recovery run verify the source and destination have
not changed. This will prevent issues on some extreme edge cases but
the main goal is for disappearing source drives with heavy damage.
e.g. A very damaged source drive disappears mid-recovery, drops off and
before would need a restart, or unplug/replug, to continue. Now we can
attempt to re-detect the drive and resume recovery without leaving the
script. If for some reason the drive order were to change then we'll
avoid using the wrong source or destination device.
