LM Demo is failing due to inconsistencies in query naming

[bharathappali]

Describe the bug
This bug was originally found by @chandrams while trying out the 0.7-rc1 build which has the changes for accelerator support for remote monitoring.

As part of that, the naming convention for gpu queries are changed from gpu to accelerator, which brings the problem in local monitoring as the metric profile for the LM are still having the older names.

sample failure:

{"message":"Exception occurred while fetching metrics from the datasource: Exception occurred while fetching metrics from the datasource: Cannot invoke \"String.replace(java.lang.CharSequence, java.lang.CharSequence)\" because \"acceleratorDetectionQuery\" is null","httpcode":400,"documentationLink":"","status":"ERROR"}

How to reproduce it
Run LM Demo with build 0.7-rc1

Expected behavior
LM Demo should run without any issue

Relevant logs

{"message":"Exception occurred while fetching metrics from the datasource: Exception occurred while fetching metrics from the datasource: Cannot invoke \"String.replace(java.lang.CharSequence, java.lang.CharSequence)\" because \"acceleratorDetectionQuery\" is null","httpcode":400,"documentationLink":"","status":"ERROR"}