Added speakers to internal_format

[vnali]

This PR is related to #61

This PR:

• Detects speakers from SRT and VTT files and keep it in internal_format.
• When the VTT is created, it checks for $internal_format['speakers'] and if there is any speaker, attaches <v {speaker_name}> to the beginning and </v> to the end of the line.
• When the SRT is created, it checks for $internal_format['speakers'] and if there is any speaker, attaches {speaker_name}: to the beginning of the line.
• Now when add() a subtitle, you can pass the speaker as keys of third param.
  ->add(0, 1, ['{speaker_name}' => 'Line 1', 'Line 2'])
->add(1, 2, ['Line 1', '{speaker_name}' =>'Line2'])
->add(2, 3, ['{speaker_name}' => 'Line 1', '{speaker_name}' =>'Line2'])
->add(3, 4, ['Line 1', 'speaker2: Line2']) // If the speaker is not available -Line1- or it is already included in the line -Line2-, you should not pass speaker separately.

[mantas-done]

Nice job, I will check this commit next week :)

[mantas-done]

Refactored code, check the newest commit from the "master" branch.

Added a new class for Webvtt.
Now you can add the speaker like this (as you suggested):
(new Webvtt)->add(0, 1, ['John' => 'hello', 'hello to you too'])->...

Also changed the name of styles (vtt_cue_settings -> settings):
(new Webvtt)->add(0, 1, 'hello', ['settings' => 'position:50% line:15% align:middle'])->...

Webtt class saves everything to the vtt key:
$internal_format[0]['vtt']

For the srt file detection of the speaker is not accurate. Checked a few files and there is text like:
These colors: blue, green and red
Which would set "These colors" as a speaker.

So for now left srt as is.

What do you think?

[vnali]

You are definitly right about false detection of SRT files, so we must skip SRT speaker detection for now, but:

• I am not sure if asking end users to use a separate class -Webvtt()- is a good idea because:
  • The end users should have a if logic to use Webvtt not Subtitle class if they work with different formats-
  • we could still use specifying speaker on add() for generating other types with known speaker format like SRT and Asking people to do (new Webvtt)->add(0, 1, ['John' => 'hello', 'hello to you too'])->content('srt') is weird. i have a feeling we should have a general add().
• although as you said, we can't have an exact speaker detection for SRT file and SRT to VTT convert, but we can still correctly detect speaker from VTT files and convert VTT to SRT formats correctly, so we could still check for speaker in internal format for generating srt content like speaker: line.
• i guess you are not happy with vtt_cue_settings on add() too, maybe we could change it to inline_cue_settings or ['vtt']['cue_settings'], ...

Please let me know if i misunderstood something
Thanks!