Can I trust testrpc spent gas amount as being close (or even exactly the same) to the actual gas amount that will be spent by my transactions when I deploy my Smart Contract into the main network?
It should be exactly the same under the same circumstances. If you get different values you should report so it can be fixed.
But currently testrpc is missing some features from byzantinum. Like REVERT opcode, which causes a big difference in the gas calculation in cases when an assert or require fails.