More BitPacking fixes (#14)

Author: gatesn

src/bitpacking.zig

@@ -60,11 +60,11 @@ test "bitpack range" {
     const fl = @import("./fastlanez.zig");
     const BP = BitPacking(fl.FastLanez(u8));
 
-    const W = 3;
+    const W = 6;
 
     var ints: [1024]u8 = undefined;
     for (0..1024) |i| {
-        ints[i] = @intCast(i % 7);
+        ints[i] = @intCast(i % std.math.maxInt(std.meta.Int(.unsigned, W)));
     }
 
     var packed_ints: [128 * W]u8 = undefined;

src/fastlanez.zig

@@ -185,8 +185,8 @@ pub fn FastLanez(comptime Element: type) type {
                     self.t += 1;
 
                     var tmp: MM1024 = undefined;
-                    if (self.input_idx == 0) {
-                        tmp = load(input, 0);
+                    if (self.bit_idx % T == 0) {
+                        tmp = load(input, self.input_idx);
                         self.input_idx += 1;
                     } else {
                         tmp = state;
@@ -197,7 +197,7 @@ pub fn FastLanez(comptime Element: type) type {
 
                     var next: MM1024 = and_rshift(tmp, shift_bits, bitmask(mask_bits));
 
-                    if (mask_bits != Width and self.input_idx < Width) {
+                    if (mask_bits != Width) {
                         tmp = load(input, self.input_idx);
                         self.input_idx += 1;