I believe the upper casing happens at the user interface level. I'll dig deeper to see if there is something in the algorithms that cares about case, but I think it was a matter of convenience in the user interface so that users didn't inadvertently enter lower and upper case and wonder why they are two different symbols.
The fact that RE discards case information for the symbols causes some problems for using some professional industry data sources. We use the in-house Reuters infrastructure as primary data source for historical, intraday and realtime data and that works just fine. The only problem is that Reuters codes are case sensitive and RE makes upper case of everything. This necessitates a double storage of symbols and a case lookup in an internal database each time RE calls up a symbol. Symbols which have the same code but different case simply cannot be used.
I realize this is not a concern for the consumer user base as most non-pro sources are upper case only, but for the pro market this makes a big difference. Many institutional sources use mixed case symbols. Any chance RE could be made to maintain the case in which the symbols were entered? If it's a matter of a backward compatibility problem, perhaps an option to make symbols case sensitive?