Once newrelic archive is unzipped, you would just need add the jvm argument to include the new relic agent or bind it to cq5 process more like (1)(1) : java -javaagent:<Path_to_newrelic.jar> -jar cq5-author-4502.jar

No, this is the correct status for reindexing and indexstats show that it is initiation phase and would be in running state and next completedwait for the reindexing to be completed and recheck these later

You'll have to check the Async IndexStats bean which would give you better idea why this is repeating ? Go to system/console/jmx > Indexstats (async) and see at the bottom of the screen

Maybe you can put the default value to <site> for path predicate at (1)(1) : http://HOST:PORT/libs/cq/core/content/tools/customsearch.html/conf/global/settings/dam/search/facets/assets/jcr:con…

Time it would take for reindexing would be dependent on the content and type of content you have. Once the reindex is complete, You can go check the indexStats Jmx mbean  in the felix console or crxde for that index and reindex=false flag would be set

Go with the we.retail implementation (1) as Geometrixx has been deprecated and structure is now based on Language masters and so on. (1) : Reusing Content: Multi Site Manager and Live Copy

You should be able to achieve this  with mix of etc mappings to map/rewrite content path and rewrite rules to handle the extensionless urls.  Have a look at (1) to get an idea on this.(1) : Extensionless URLs with Adobe Experience Manager | Olson Digital

You'll have to adjust the xss config to avoid the antisamy configuration to strip off these attributes.Once you click save, Check the error.log that would provide the message that would give you an idea that antisamy is at play.To fix this: You can follow (1) on similar lines.(1) : XSS Filter issue ...

Can you enable debug logs for the search and help me with the query that is generated at the end? Maybe we can somehow tune the index definitions and make it better performing ?

This configuration of max 5 version and max age 30 related to to version purge i.e deleting versions which are not in use. Only if you run or  schedule a maintenance task to purge versions this would come into play.There is no finite limit on how many versions you can create in the system.