Feature request: reusable bias

[HanatoK]

It should be possible to use the bias energy, or other output components, of a colvarbias instance as the inputs of another colvarbias instance. For example, a histogram bias can reuse the bias energies from other biases to perform reweighting on-the-fly.

[giacomofiorin]

As commented in #709 we have already several cases where we are running all biases in sequence:

colvars/src/colvarmodule.cpp

Lines 1016 to 1042 in a587282

	if (proxy->check_smp_enabled() == COLVARS_OK && !biases_need_main_thread) {
	if (use_scripted_forces && !scripting_after_biases) {
	// calculate biases and scripted forces in parallel
	error_code |= proxy->smp_biases_script_loop();
	} else {
	// calculate biases in parallel
	error_code |= proxy->smp_biases_loop();
	}
	} else {
	if (use_scripted_forces && !scripting_after_biases) {
	error_code |= calc_scripted_forces();
	}
	// Straight loop over biases on a single thread
	cvm::increase_depth();
	for (bi = biases_active()->begin(); bi != biases_active()->end(); bi++) {
	error_code |= (*bi)->update();
	if (cvm::get_error()) {
	cvm::decrease_depth();
	return error_code;
	}
	}
	cvm::decrease_depth();
	}

So if there is a dependency like the one you describe, we could just do the same (until we have the AST that you proposed elsewhere).

The only tricky detail would be ensuring that the biases are listed internally in the same order as they were defined in the user input. colvarmodule::parse_biases() reads all biases of the same type at once, and goes alphabetically over bias types. However, we could record the position of each bias in a new input string and use that position to sort all newly added biases?