cpu/esp32: use ESP-IDF i2c HAL for ESP32 periph/i2c

[gschorcht]

Contribution description

This PR is a split-off from PR #17841 and provides the following changes:

• ESP-IDF i2c HAL and ESP-IDF interrupt HAL are used in hardware implementation (module esp_i2c_hw).
• In software implementation, which is used by default, fixed delay values are replaced by calculated delays measured in CPU cycles. The advantage is that for each ESP SoC only a clock calibration offset has to be specified. The delay measured in CPU cycles are then derived from current CPU frequency for the given bus speed. The disadvantage is that the calculated delays are not as precise as the predefined fixed delays before.

This PR depends on PR #18270.

Testing procedure

1. Green CI (compilation of hardware implementation is covered by board esp32-ci).
2. Compile and test the hardware implementation with

   USEMODULE='esp_i2c_hw esp_log_startup' BOARD=esp32-wroom-32 make -j8 -C tests/periph_i2c flash term
   

3. Compile and test the software implementation with

   USEMODULE='esp_log_startup' BOARD=esp32-wroom-32 make -j8 -C tests/periph_i2c flash term
   

Issues/PRs references

Split-off from PR #17841
Depends on PR #18270

[benpicco]

Suggested change

	uint8_t module; /**< I2C module identifier */
	i2c_speed_t speed; /**< I2C bus speed */
	gpio_t scl; /**< GPIO used as SCL pin */
	gpio_t sda; /**< GPIO used as SDA pin */
	bool scl_pullup; /**< Pullup enabled for SCL pin */
	bool sda_pullup; /**< Pullup enabled for SDA pin */
	gpio_t scl; /**< GPIO used as SCL pin */
	gpio_t sda; /**< GPIO used as SDA pin */
	uint8_t module; /**< I2C module identifier */
	i2c_speed_t speed; /**< I2C bus speed */
	bool scl_pullup; /**< Pullup enabled for SCL pin */
	bool sda_pullup; /**< Pullup enabled for SDA pin */

struct packing

[gschorcht]

It seems that the reordering does not save any byte. Should I change it anyway?

[benpicco]

Hm I had hoped this might save a few bytes of ROM, but it's no big deal either way

[benpicco]

Please squash

[gschorcht]

@benpicco Thanks for reviewing and merging

[gschorcht]

@benpicco This was the last PR that had to be merged to continue the work. Many thanks again for the quick review and merge of all these split-offs from PR #17841, I really appreciate it. Thanks to your reviews we were able to merge them all into the release, so we should have a consistent state of the APIs used. The only remaining PR #18260 is completely independent of the other changes in PR #17841 and can be merged later, even after the changes in PR #17842 and in conjunction with PR #17844.