Don't sync SOF between device and host

This makes it so that the SOF clock on the downstream facing side is
no longer synchronized to the upstream facing side on the single Pico
variant.

While I think overall the behavior was desirable, unfortunately it
caused issues in certain situations, causing devices to not work if
they were plugged into HID Remapper before it was plugged into the PC
or if the PC was turned on with everything already plugged in. There
were related issues with KVM switches and sleep/wake. Also because the
processing engine didn't run at all when there were no SOF packets
coming from the host, HID Remapper couldn't be used as a USB-to-GPIO
decoder without a host.

All of these issues could probably be addressed while keeping the SOF
synchronization with some additional effort. We might revisit this in
the future.

Author: jfedor2

firmware/src/remapper_single.cc

@@ -11,11 +11,20 @@
 #include "remapper.h"
 #include "tick.h"
 
+static bool __no_inline_not_in_flash_func(manual_sof)(repeating_timer_t* rt) {
+    pio_usb_host_frame();
+    set_tick_pending();
+    return true;
+}
+
+static repeating_timer_t sof_timer;
+
 void extra_init() {
     pio_usb_configuration_t pio_cfg = PIO_USB_DEFAULT_CONFIG;
     pio_cfg.pin_dp = PICO_DEFAULT_PIO_USB_DP_PIN;
     pio_cfg.skip_alarm_pool = true;
     tuh_configure(BOARD_TUH_RHPORT, TUH_CFGID_RPI_PIO_USB_CONFIGURATION, &pio_cfg);
+    add_repeating_timer_us(-1000, manual_sof, NULL, &sof_timer);
 }
 
 uint32_t get_gpio_valid_pins_mask() {
@@ -123,14 +132,7 @@ void send_out_report() {
     do_send_out_report();
 }
 
-static int64_t __no_inline_not_in_flash_func(manual_sof)(alarm_id_t id, void* user_data) {
-    pio_usb_host_frame();
-    set_tick_pending();
-    return 0;
-}
-
 void __no_inline_not_in_flash_func(sof_callback)() {
-    add_alarm_in_us(150, manual_sof, NULL, true);
 }
 
 void get_report_cb(uint8_t dev_addr, uint8_t interface, uint8_t report_id, uint8_t report_type, uint8_t* report, uint16_t len) {