Collect docs metadata on prepare

This script is run when locally installing the package, see
https://docs.npmjs.com/cli/v6/using-npm/scripts#life-cycle-scripts.
So it is more likely that the static/docs-metadata.json file exists when
we do invoke any of the other commands. Note that this is required
becaus the docusaurus.config.js does load this file as well.

docs/package.json

@@ -4,11 +4,12 @@
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",
-    "start": "yarn enrich-document-metadata && docusaurus start",
+    "prepare": "yarn enrich-document-metadata",
+    "build": "yarn prepare && docusaurus build",
+    "start": "docusaurus start",
     "validate:inputs": "./validate-api.js publish '/' '../hydra-node/golden/ReasonablySized (ClientInput (Tx BabbageEra)).json'",
     "validate:outputs": "./validate-api.js subscribe '/' '../hydra-node/golden/ReasonablySized (TimedServerOutput (Tx BabbageEra)).json'",
     "validate": "yarn validate:inputs && yarn validate:outputs",
-    "build": "yarn enrich-document-metadata && docusaurus build",
     "swizzle": "docusaurus swizzle",
     "deploy": "docusaurus deploy",
     "clear": "docusaurus clear",