Caliper bindings are broken

[jngrad]

The Caliper bindings that mark the start and end of a portion of code to instrument are incorrect:

espresso/src/profiler/include/profiler/profiler.hpp

Lines 62 to 64 in 09c6958

	inline void begin_section(const std::string &name) {
	ESPRESSO_PROFILER_MARK_BEGIN(name.c_str());
	}

espresso/src/profiler/include/profiler/profiler.hpp

Lines 71 to 73 in 09c6958

	inline void end_section(const std::string &name) {
	ESPRESSO_PROFILER_MARK_BEGIN(name.c_str());
	}

The second one should be ESPRESSO_PROFILER_MARK_END. The regression was introduced in 1df1009. When calling these functions inside the body of a Python loop, the Caliper summary creates a recursive data structure for the marked region, since the end marker is missing.

The Cython interface passes Python string objects to C++ functions that take std::string as arguments, triggering runtime exceptions. The interface should use espressomd.utils.to_char_pointer() to handle type conversion. The regression was introduced in df5d9d2.

There are no test cases for this feature.