Fix param 4 and 5 for GET_NTH_CLOSEST_VEHICLE_NODE_ID_WITH_HEADING

[CreepyQuietWeirdo]

See Freemode where the 5th argument being used, var4, couldn't be a Vector3, but must be float, and Var5 declared directly below it is being used as an int. These variables shouldn't overlap. Freemode @v3788:

int func_1119(Vector3* vParam0, Vector3* vParam1, var* uParam2)//Position - 0x8376F
{
[...]
	Vector3 vVar4; // <-- Impossible, can't be a struct size  of 3 with iVar5 following it immediately
	int iVar5;
[...]
	iVar5 = 0;
[...]
		iVar5 += 2;
		iVar5++;
[...]
	iVar5 += 4;
[...]
		iVar8 = PATHFIND::GET_NTH_CLOSEST_VEHICLE_NODE_ID_WITH_HEADING(*vParam0, (iVar0 * iVar17), &vVar4, &fVar9, iVar5, fVar13, fVar14);

If you reverse the native, it is clear that this parameter is filled with the heading in 0.0 to 360.0 degrees as a float pointer, not as a Vector3 pointer. (If it were, a wrapper would also be present to convert from 3 consecutive floats to 3 script floats with 8-byte alignment and there is none.) Furthermore the 6th argument is used internally as an int pointer, for numLanes. It seems the only difference between this function and GET_NTH_CLOSEST_VEHICLE_NODE_WITH_HEADING is that the Vector3* outHeading is omitted, instead of the numLanes parameter.

This also explains why in Freemode func_1119 v3788 PATHFIND::GET_VEHICLE_NODE_POSITION is called immediately after calling PATHFIND::GET_NTH_CLOSEST_VEHICLE_NODE_ID_WITH_HEADING, this wouldn't make sense otherwise.

[alloc8or]

Looks like someone fixed this for rdr3-nativedb-data already: https://alloc8or.re/rdr3/nativedb/?n=0x4114EAA8A7F7766D
Maybe we should just copy that signature for consistency?