fix: only prepare the data once

[SamuelLarkin]

PR Goal?

This PR is required to fix EveryVoiceTTS/EveryVoice#419

Fixes?

Doesn't fix EveryVoiceTTS/EveryVoice#419 but is required

Feedback sought?

Ready to merge

Priority?

part of alpha, so may be medium-high

Tests added?

No, but may be I should add a small test in EV.

How to test?

for max_epochs in 3 6 9 12 15; do
   echo "Training up to $max_epochs"
   srun everyvoice train text-to-spec \
      config/everyvoice-text-to-spec.yaml \
      --config-args training.finetune_checkpoint=$PWD/logs_and_checkpoints/FeaturePredictionExperiment/base/checkpoints/last.ckpt \
      --config-args training.max_epochs=$max_epochs
done

Confidence?

Good

Version change?

No

Related PRs?

EveryVoiceTTS/EveryVoice#473