add manual training data for common questions

[kurtismgit]

NLC using the crawler generated training data answers several of these questions well, but it wouldn't hurt to explicitly add manual training statements in addition to the crawler generated statements for questions that might be common for suggest services. Here's a starting list:

@ibmcloudbot suggest service to translate languages
@ibmcloudbot suggest service to classify SMS messages
@ibmcloudbot suggest service to integrate weather data
@ibmcloudbot suggest a service to record phone calls
@ibmcloudbot suggest services for caching
@ibmcloudbot suggest services to migrate data from DB2
@ibmcloudbot suggest services for analytics
@ibmcloudbot suggest services to integrate with watson
@ibmcloudbot suggest services for online banking
@ibmcloudbot suggest services to host databases on the cloud
@hubot suggest services for devops in the cloud
@hubot suggest services to perform application deployments
suggest service to build mobile apps on the cloud
suggest services for mobile app developers