Fix leaked Path class

[bparks13]

Somewhere in creating the OnixSourceCanvas, specifically the device interfaces for Neuropixels 1.0f (i.e., Bno055Interface and NeuropixV1Interface), there is a leaked Path. Initial testing also indicated that there are device classes leaked from these interfaces. Figure out where the issue is coming from, and plug the leak.

See #11 for a related issue.

[bparks13]

In #17, adding the related macros shows that the devices instantiated are actually what is leaking, but I can't figure out exactly where they are leaking. There are no additional classes that are registered as being leaked, indicating it is something with the device pointers themselves. Specifically, the following lines can be removed to fix the leak, but then nothing downstream happens:

onix-source/Source/OnixSource.cpp

Lines 169 to 243 in d9d4918

	if (devices[dev_idx].id == ONIX_NEUROPIX1R0)
	{
	auto np1 = std::make_unique<Neuropixels_1>("Probe-" + String::charToString(probeLetters[npxProbeIdx]), this, devices[dev_idx].idx, ctx);
	int res = np1->enableDevice();
	if (res != 0)
	{
	if (res == -1)
	{
	LOGE("Device Idx: ", devices[dev_idx].idx, " Unable to read probe serial number. Device not found.");
	continue;
	}
	else if (res == -2)
	{
	LOGE("Device Idx: ", devices[dev_idx].idx, " Error enabling device stream.");
	continue;
	}
	else if (res == -3 || res == -4)
	{
	LOGE("Missing or invalid calibration file(s). Ensure that all calibration files exist and the file paths are correct.");
	}
	}
	sources.add(np1.release());
	npxProbeIdx++;
	}
	else if (devices[dev_idx].id == ONIX_BNO055)
	{
	auto bno = std::make_unique<Bno055>("BNO-" + String::charToString(probeLetters[bnoIdx]), devices[dev_idx].idx, ctx);
	int result = bno->enableDevice();
	if (result != 0)
	{
	LOGE("Device Idx: ", devices[dev_idx].idx, " Error enabling device stream.");
	continue;
	}
	sources.add(bno.release());
	bnoIdx++;
	}
	else if (devices[dev_idx].id == ONIX_DS90UB9RAW)
	{
	LOGD("Passthrough device detected");
	//initialize main i2c parameter
	auto serializer = std::make_unique<I2CRegisterContext>(DS90UB9x::SER_ADDR, devices[dev_idx].idx, ctx);
	serializer->WriteByte((uint32_t)DS90UB9x::DS90UB9xSerializerI2CRegister::SCLHIGH, 20);
	serializer->WriteByte((uint32_t)DS90UB9x::DS90UB9xSerializerI2CRegister::SCLLOW, 20);
	auto EEPROM = std::make_unique<HeadStageEEPROM>(devices[dev_idx].idx, ctx);
	uint32_t hsid = EEPROM->GetHeadStageID();
	LOGD("Detected headstage ", hsid);
	if (hsid == 8) //Npix2.0e headstage, constant needs to be added to onix.h
	{
	auto np2 = std::make_unique<Neuropixels2e>("Probe-" + String::charToString(probeLetters[npxProbeIdx]), devices[dev_idx].idx, ctx);
	int res = np2->enableDevice();
	if (res != 0)
	{
	if (res == -1)
	{
	LOGE("Device Idx: ", devices[dev_idx].idx, " Unable to read probe serial number. Device not found.");
	}
	//TODO add other errors if needed
	continue;
	}
	npxProbeIdx += np2->getNumProbes();
	sources.add(np2.release());
	}
	}

If these lines are left in, I have tried to selectively remove lines from other classes to find the leak, such as removing the lines that create the interfaces in OnixSourceCanvas, but none of these modifications fix it.

Prior to f8d65b2, I don't remember this being an issue. However, cherry-picking changes from this commit do not seem to fix the issue, so maybe it started occurring prior to this and I did not notice.