Switch Travis config to use `arm64` architecture by default

[roborourke]

Since Local Server supports Arm builds now we can switch Travis to use Arm chips by default. This should have the following benefits:

• Usually smaller docker image sizes
• Faster execution time

We should start with an experiment and benchmark some existing module tests e.g. for this repo and see what difference using arm makes.

Acceptance criteria:

• All Altis modules use Arm architecture on travis by default
• The project level config uses Arm by default
• The architecture is documented along with how to modify it - https://github.com/humanmade/altis-dev-tools/tree/master/travis
• The expected speed improvements are documented

For Altis Team Use

Ready for Work Checklist

Is this ticket ready to be worked on? See
the Play Book Definition of Ready

• Is the title clear?
• Is the description clear and detailed enough?
• Are acceptance criteria listed?
• Have any dependencies been identified? (Optional)
• Have any documentation/playbook changes been identified? (Optional)
• Is an estimate or time box assigned?
• Is a priority label assigned?
• Is this ticket added to a milestone?
• Is this ticket added to an epic? (Optional)

Completion Checklist

Is this ticket done? See
the Play Book Definition of Done

• Has the acceptance criteria been met?
• Is the documentation updated (including README)?
• Do any code/documentation changes meet project standards?
• Are automatic tests in place to verify the fix or new functionality?
  • Or are manual tests documented (at least on this ticket)?
• Are any Playbook/Handbook pages updated?
• Has a new module release (patch/minor) been created/scheduled?
• Have the appropriate backport labels been added to the PR?
• Is there a roll-out (and roll-back) plan if required?

[mikelittle]

Added an extra acceptance criterion. It would be good to note the speed improvement (and maybe the size improvement too) as we are looking at the difference.

[wisyhambolu]

Hey team! Please add your planning poker estimate with Zenhub @jerico @kovshenin @mikelittle