tools: Refactor notification API

[theotherjimmy]

Description

Historically, the notification API is coupled to an mbedToolchain instance.
This is problematic for things that could use notification without a toolchain
object, such a export zipping, or post build merge scripts, and this is a poor
coupling of the API that requires passing many, many parameters through to the
toolchain, such as verbose and silent.

This PR decouples the notification API from the toolchain object, and delegates
implementaion of this Notifier API to a single implementation:
TerminalNotifier. In testing this PR I prototyped another implementation as
part of the Online Compiler's build back end.

This new API allowed me to decouple the verbosity parameter handling, -v and
--silent, from the build API entirely. They are now completely oblivious to
these parameters, instead calling methods on the Notifier object that they
are passed.

Pull request type

[ ] Fix
[x] Refactor
[ ] New target
[ ] Feature
[ ] Breaking change

[theotherjimmy]

TODO for tomorrow theotherjimmy: Fix those dang tools tests.

[theotherjimmy]

/morph build

[mbed-ci]

Build : ABORTED

Build number : 1884
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/6781/

[0xc0170]

/morph build

[mbed-ci]

Build : ABORTED

Build number : 1891
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/6781/

[0xc0170]

Aborted again, but not by us, tests are failing and then results in timeout? Please review

[theotherjimmy]

@0xc0170 Yeah, it looks like IAR everything is failing for some targets.

[theotherjimmy]

@0xc0170 I got the bug, I think. The test builds were sharing a notifier, which basically squared the amount of output we got.

[theotherjimmy]

/morph build

[mbed-ci]

Build : ABORTED

Build number : 1910
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/6781/

[theotherjimmy]

/morph build

[theotherjimmy]

So the prior abort was caused by a combination of a few things leading up to the test builder consuming an unbounded (and very, very large) amount of RAM. That commit seems to keep it capped at a little less than 400MB on my machine. At particularly bad spots, the prior commit could consume >2GB RAM on my machine.

[theotherjimmy]

/morph build

[mbed-ci]

Build : SUCCESS

Build number : 1918
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/6781/

Triggering tests

/morph test
/morph uvisor-test
/morph export-build
/morph mbed2-build

[mbed-ci]

Exporter Build : SUCCESS

Build number : 1563
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/exporter/6781/

[0xc0170]

@theotherjimmy Can you review the jenkins CI failure, there's a build error

[theotherjimmy]

@0xc0170 Jenkins passed.

[mbed-ci]

Test : FAILURE

Build number : 1734
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/6781/1734

[theotherjimmy]

/morph test

[mbed-ci]

Test : SUCCESS

Build number : 1740
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/6781/1740

[screamerbg]

Quite a large change set, but for all the important things. Thanks, @theotherjimmy!

[jeromecoutant]

Hi @theotherjimmy

Since this PR merge, build python script is not working any more :-(

$ python tools/build.py -m NUCLEO_F030R8 -t ARM
name 'TerminalNotifer' is not defined

[0xc0170]

@jeromecoutant Thanks for reporting. we will look at this