[tokenize] Avoid istringstream in delimiter-based tokenize

[Mohxen]

Problem

tokenize(str, token) currently constructs a std::istringstream for every call.
This is more expensive than needed for splitting a string on a single delimiter,
because it brings in iostream and locale machinery.

This helper is used in SWSS hot paths, including route processing in orchagent.
During route burst profiling, perf showed time in:

• swss::tokenize
• std::getline
• std::basic_ios::_M_cache_locale
• std::locale
• std::vector<std::string> growth
• operator new

Fix

Replace the istringstream/getline implementation with direct
std::string::find() / substr() delimiter scanning.

This matches the implementation style already used by the limited-token tokenize
overload added in #96, while preserving the observed behavior of the base overload.

Testing

Configured locally with YANG modules disabled:

./autogen.sh
./configure --enable-debug --disable-yangmodules 'CXXFLAGS=-O0 -g'

[linux-foundation-easycla]

• ❌ - login: @Mohxen / name: Mohxen. The commit (26edfdb) is not authorized under a signed CLA. Please click here to be authorized. For further assistance with EasyCLA, please visit our EasyCLA portal and chat with our support bot.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).