[SYCL] Revert support for templated call operators

[srividya-sundaram]

This PR reverts the support for templated call operators in functors and functors without call operators which were introduced in #7104 and #7970

We found a regression internally with #7104 that #7970 attempted to fix. A subsequent review of #7970 internally, lead to identifying some pre-existing caveats with regards to supporting templated call operators in functor, supporting cases where there are multiple call operators in functor and making sure the right instantiated version is selected etc.

In order to better address the pre-existing issues as well as gaps in #7970 , it was decided to revert the above two
PRs.

[smanna12]

LGTM. Can you please add some description why you are reverting the patch for future reference?

[premanandrao]

LGTM. Can you please add some description why you are reverting the patch for future reference?

We found a regression internally with #7104 that #7970 attempted to fix. A subsequent review of #7970 internally, found some deficiencies with the patch (with templated call operators). Rather than rush a fix for that, @srividya-sundaram is backing out both PRs and intends to architect a solution that addresses the concerns raised. She may be able to provide a better message to add in the description.

[smanna12]

LGTM. Can you please add some description why you are reverting the patch for future reference?

We found a regression internally with #7104 that #7970 attempted to fix. A subsequent review of #7970 internally, found some deficiencies with the patch (with templated call operators). Rather than rush a fix for that, @srividya-sundaram is backing out both PRs and intends to architect a solution that addresses the concerns raised. She may be able to provide a better message to add in the description.

Thanks @premanandrao for the explanation!

[srividya-sundaram]

LGTM. Can you please add some description why you are reverting the patch for future reference?

Updated the PR description with details.