Car Tweaks

[larsll]

Issue #, if available:

Description of changes:
These fixes will make the following changes to the cars:

• DREM optimizes the Tensorflow models to OpenVINO; this patch will now enable the car to take benefit, reducing model load times down to 1-2 seconds. This patch is equal to Optimization Caching aws-deepracer/aws-deepracer-model-optimizer-pkg#2 and builds on Model Optimizer Lamda Function #9.
• A small CSS change prevents the UI to zoom in on a double-click on touch-screens.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[larsll]

@marcobuss @esbjj @StevenAskwith I would strongly suggest to merge this in before re:Invent; the model load times will simplify life a lot, and enable more efficient model changes during 2 minute runs.

[StevenAskwith]

@larsll most of the team in on holiday this week, will review next week, but the idea looks good!

[MarkRoss-Eviden]

Hi @StevenAskwith FYI as part of clarification questions for the finals we've asked if this functionality will be available. As you can imagine with limited practice (2 hour slot for 16 competitors on the Tues and Weds before round 1) and the limited time per racer in all rounds (2 min attempts) it's quite material whether changing models mid run takes ~45s or 2s.

[StevenAskwith]

@MarkRoss-Eviden I triggered the deploy to my test environment ~30 mins ago, will update after I've tested.