New features for NLP, speech-to-Text and text-to-speech

• Improved error handling for vectorised gl_nlp() (#55)
• gl_nlp()'s classifyText returns list of data.frames, not data.frame
• Fix gl_nlp when nlp_type='classifyText'
• customConfig available for gl_speech
• Add support for SSML for gl_talk() (#66)
• Add support for device profiles for gl_talk() (#67)
• Add support for tuneR wave objects in gl_speech() - (#62 thanks @muschellij2)
• Add check for file size for audio source - (#62 thanks @muschellij2)