This update for tesseract-ocr fixes the following issues:

Update to 5.3.4

- Fixes for autoconf, clang and sw builds
- Send output of combine_tessdata -d to stdout instead of stderr.
Fixes #4149 in #4150
- Move bail_out function before libtoolize check in #4151
- Improve OCR for an image URL
- Fail on curl download errors
- Add new parameter curl_cookiefile for curl_easy_setopt in #4156
- Set User-Agent: header field in HTTP request for curl downloads
- Force TCP v4 for socket to ScrollView server. Fixes #3000 in #4162
- Fix some compiler warnings and avoid unnecessary conversions
from std::string to char pointer in #4174
- Fix a tiny typo in publictypes.h in #4178
- Fixes for autoconf, clang and sw builds
- Other small improvements for code and documentation.

Update to 5.3.3

- Disable -mfpu=neon for aarch64 in #4098
- Fix build without git clone in cloned directory in #4099
- Fix some issues which were reported by Coverity Scan in #4097
- Update ScrollView.java in #4103
- Fix some code comments in #4113
- Optimize function ImageFind::FindImages in #4114
- Rename BibTex file to please GitHub in #4115
- Fix Broken URLs in citations.bib in #4118
- initDSProfile: correct std::vector usage in #4124
- Fix typo in stepblob.h in #4133
- Fix regression in layout detection since 5.0.0 (fixes issue #4014) in #4136
- Update ScrollView.java in #4104
- Fix loading of sublangs (regression) in #4141

Update to 5.3.2

- fix: Fix snap package building in #4043
- Support for Sgaw and W Pwo Karen languages in the Myanmar validator. in #4065
- Replace bool array by more compact vector in #4067
- Replace deprecated sprintf in #4068
- Improve format of logging from lstmtraining in #4066
- Clean code in #4071
- Abort with error message if OSD is requested with LSTM-only model in #4073
- Fix typos in #4096