optimize sequence readers and patch a bug in the format guesser

Former-commit-id: 9dce1e96c57ae9a88c26fac5c8e1bdcdc2c0c7a5

pkg/obiformats/fastqseq_read.go

@@ -99,7 +99,11 @@ func FastqChunkReader(r io.Reader, size int) (chan FastxChunk, error) {
 	out := make(chan FastxChunk)
 	buff := make([]byte, size)
 
-	n, err := r.Read(buff)
+	n, err := io.ReadFull(r, buff)
+
+	if err == io.ErrUnexpectedEOF {
+		err = nil
+	}
 
 	if n > 0 && err == nil {
 		if n < size {
@@ -130,13 +134,19 @@ func FastqChunkReader(r io.Reader, size int) (chan FastxChunk, error) {
 						index: idx,
 					}
 					idx++
+				} else {
+					size = size * 2
 				}
 
 				buff = slices.Grow(buff[:0], size)[0:size]
-				n, err = r.Read(buff)
+				n, err = io.ReadFull(r, buff)
 				if n < size {
 					buff = buff[:n]
 				}
+
+				if err == io.ErrUnexpectedEOF {
+					err = nil
+				}
 				// fmt.Printf("n = %d, err = %v\n", n, err)
 			}