Three small changes to compile (again) under C++98 (closes #1192)

[eddelbuettel]

PRs #1184 and #1187 added support for (function) 'signatures' to generate R function interfaces. The string processing in there (inadvertendly) used C++11 idioms in the three spots. We all missed this during code review.

This should (in general) be a non-issue as R itself now imposes C++11 (in R 4.0.) and even C++14 (in R 4.1.). Of course this is both dependent on the compiler, and can be overridden. While it generally works (in all our testing) we saw in #1192 that it can fail.

So while we "probably" should just move to C++11 (which should not have side effects from the compilation of the package on to other packages), it is easy enough to restore plain old C++98 compilation -- so we may as well (while we discuss / test moving to C++11). This PR does the (simple) restoration of C++98 buildability.

Checklist

• Code compiles correctly
• R CMD check still passes all tests
• Prefereably, new tests were added which fail without the change
• Document the changes by file in ChangeLog

[Enchufa2]

LGTM!