[llvm] Fix to generalize the canSkipClobberingStore (PR #174137)

[via llvm-commits]

Bhuvan1527 wrote:

> This is very obviously incorrect. Please read the comment above the function and try to understand what this code is actually doing. Then look at the code in the issue and try to understand why it's valid to optimize that code.
> 
> (Hint: Everything here hinges on "writes the same value that MemLoc contains now". The value being an argument tells you absolutely nothing about the current value of the location.)

Hi @nikic 

Sorry for leaving this pr untouched. I got occupied with my semester works. 

Considering your comment. I've approached this issue as follows : **skipping the store if it is storing same value loaded  via loaded instruction.**  

I've updated the pr. Also had to re-written the existing code to fit the new code changes. 

looking forward for your comments.
Thanks. 

https://github.com/llvm/llvm-project/pull/174137